Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Algorithms for automated measurements of the radioresponse of live cells at therapeutic doses Spadinger, Ingrid Teresa 1990

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1990_A1 S69.pdf [ 14.53MB ]
Metadata
JSON: 831-1.0084995.json
JSON-LD: 831-1.0084995-ld.json
RDF/XML (Pretty): 831-1.0084995-rdf.xml
RDF/JSON: 831-1.0084995-rdf.json
Turtle: 831-1.0084995-turtle.txt
N-Triples: 831-1.0084995-rdf-ntriples.txt
Original Record: 831-1.0084995-source.json
Full Text
831-1.0084995-fulltext.txt
Citation
831-1.0084995.ris

Full Text

ALGORITHMS FOR AUTOMATED MEASUREMENTS OF THE RADIORESPONSE OF LIVE CELLS AT THERAPEUTIC DOSES by INGRID TERESA SPADINGER B.A.Sc., The U n i v e r s i t y of B r i t i s h Columbia, 1984 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY i n THE FACULTY OF GRADUATE STUDIES Department of Physics We accept t h i s thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA February 1990 © Ingrid Teresa Spadinger, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of P H Y S 1 C-S The University of British Columbia Vancouver, Canada Date ftUCr<4rt ^ mO DE-6 (2/88) ABSTRACT The s t u d y o f t h e e f f e c t s o f damaging a g e n t s on l i v i n g c e l l s i s o f i m p o r t a n c e b o t h f o r t h e assessment o f t h e p o t e n t i a l consequences o f e x p o s u r e , and f o r t h e advancement o f m e d i c a l t e c h n o l o g i e s c o n c e r n e d w i t h t h e t r e a t m e n t o f c a n c e r and o t h e r d i s e a s e s . Such a g e n t s , however, a r e commonly s t u d i e d a t doses and l e v e l s o f e f f e c t much h i g h e r t h a n t h o s e seen i n e n v i r o n m e n t a l o r even m e d i c a l e x p o s u r e s . S t u d i e s o f c e l l s u r v i v a l a f t e r t r e a t m e n t w i t h i o n i z i n g r a d i a t i o n , f o r i n s t a n c e , g e n e r a l l y i n v o l v e the measurement o f d o s e - r e s p o n s e o v e r s e v e r a l decades o f c e l l k i l l on a l o g a r i t h m i c s c a l e . I n c o n t r a s t , c e l l s u r v i v a l r a t e s f o r a t y p i c a l c l i n i c a l t r e a t m e n t dose a r e on the o r d e r o f 50%. S u r v i v a l measurements i n t h i s f i r s t decade o f c e l l k i l l r e q u i r e t h a t the e x a c t f a t e o f thousands o f c e l l s i s d e t e r m i n e d i n a s i n g l e e x p e r i m e n t . T h i s i s beyond human c a p a b i l i t i e s . Measurements a t t h e s e l e v e l s o f e f f e c t t h e r e f o r e r e q u i r e t h e development o f a r a p i d , automated sy s t e m o f c e l l d e t e c t i o n , c h a r a c t e r i z a t i o n , and f o l l o w -up. The p r i m a r y aim o f t h i s t h e s i s was t o t e s t t he h y p o t h e s i s t h a t the a l g o r i t h m s n e c e s s a r y f o r g e n e r a t i n g s u r v i v a l d a t a w i t h s u c h an automated sys t e m c a n be d e v e l o p e d t o s t a n d a r d s o f a c c u r a c y comparable t o t h o s e o f an e x p e r i e n c e d human o b s e r v e r . U s i n g an image c y t o m e t r y d e v i c e s p e c i f i c a l l y d e s i g n e d f o r t h e d e t e c t i o n and a n a l y s i s o f l i v e , u n s t a i n e d c e l l s , automated s c a n n i n g p r o c e d u r e s were o p t i m i z e d f o r s e l e c t e d c e l l l i n e s , and means o f m a i n t a i n i n g a p p r o p r i a t e f o c u s l e v e l s d u r i n g a s c a n were d e v i s e d . A l g o r i t h m s t o d i s t i n g u i s h c e l l s from o t h e r o b j e c t s d e t e c t e d i n t h e f l a s k were a l s o d e v e l o p e d . These p e r f o r m e d w i t h comparable a c c u r a c y , b u t a t g r e a t e r speed, t h a n c o u l d be a c h i e v e d by a human o b s e r v e r . i i i Because the endpoint used i n s u r v i v a l measurements i s the a b i l i t y of t r e a t e d c e l l s to p r o l i f e r a t e to form c o l o n i e s , the hypothesis that automated methods of assaying colony formation could be developed was a l s o t e s t e d . Using r a p i d l y c o l l e c t e d , low r e s o l u t i o n image data obtained at l o c a t i o n s i n the t i s s u e c u l t u r e v e s s e l where i n d i v i d u a l c e l l s had been detected on the day of treatment, i t was found that the s u r v i v a l s t a t u s at 70-90% of these l o c a t i o n s could be determined a u t o m a t i c a l l y . Manual assessments were r e q u i r e d at the remaining l o c a t i o n s . The f i n a l o b j e c t i v e of t h i s t h e s i s was to use the methods developed f o r automated sample s e l e c t i o n and s u r v i v a l assessment to examine a question r e l e v a n t to both r a d i o t h e r a p e u t i c a p p l i c a t i o n s and the understanding of mechanisms of r a d i a t i o n a c t i o n . In p a r t i c u l a r , the Theory of Dual R a d i a t i o n A c t i o n ( K e l l e r e r and R o s s i , 1972), which p r e d i c t s that the r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s (RBE) of d i f f e r e n t m o d a l i t i e s of s p a r s e l y i o n i z i n g r a d i a t i o n s may change d r a m a t i c a l l y w i t h decreasing dose, was t e s t e d experimentally. Using two d i f f e r e n t mammalian c e l l l i n e s , the experimental data i n general r e j e c t e d the p r e d i c t i o n s of t h i s theory, although a modified theory, developed by the same authors, could accommodate the r e s u l t s . In summary, the data showed that there may be a s l i g h t dose dependence i n the RBE of the r a d i a t i o n s t e s t e d . S p e c i f i c a l l y , the RBE of low energy X-rays r e l a t i v e to 6 0 C o 7-rays was found to increase s l i g h t l y w i t h decreasing dose. The measured RBE's i n the zero-dose l i m i t were -1.4 f o r 55 kVp X-rays and -1.1-1.2 f o r 250 kV^ X-rays. High energy (11 MeV) e l e c t r o n s , on the other hand, showed a small decrease i n RBE r e l a t i v e to 6 0 C o 7-rays as the dose approached zero, having a zero-dose l i m i t of -0.95. A l l of the aforementioned r a d i a t i o n m o d a l i t i e s had an RBE of 1.0-1.1 at h i g h doses (10 Gy or more). i v TABLE OF CONTENTS Page ABSTRACT i i TABLE OF CONTENTS i v LIST OF TABLES v i i i LIST OF FIGURES i x LIST OF ABBREVIATIONS x i i ACKNOWLEDGEMENTS x i i i 1. INTRODUCTION 1.1 Measurement of the E f f e c t s of Damaging Agents on C e l l s 1.1.1 Reasons for Measuring Damage to Individual C e l l s 1 1.1.2 Endpoints Used to Quantify C e l l u l a r Damage 2 1.1.3 Data Requirements and Techniques Employed i n C e l l S urvival Studies 3 1.1.4 Applications of Image Cytometry to C e l l S u r v i v a l Studies 10 1.2 Ionizing Radiation as a Damaging Agent 1.2.1 Properties of Ionizing Radiation and D e f i n i t i o n of Radiation Dose 11 1.2.2 Action of Radiation on L i v i n g C e l l s 16 1.2.3 Importance of Studying Low Dose Radiation E f f e c t s 19 1.3 Thesis Objectives 22 2. CELL HANDLING AND PREPARATION 2.1 V79 C e l l Culture 24 2.2 CHO C e l l Culture 25 2.3 Preparation of Flasks for the C e l l Analyzer 26 V 3. THE CELL ANALYZER 3.1 Important Considerations Influencing the System Design 27 3.2 Image Detection 28 3.3 Image Generation 3.3.1 Image Formation i n the Light Microscope 30 3.3.2 Imaging of Phase Objects Using S p a t i a l F i l t e r i n g Techniques 36 3.3.3 Evaluation of S p a t i a l F i l t e r i n g Methods for C e l l Detection Purposes 42 3.3.4 Generation of C e l l Signals v i a the "Lens" E f f e c t 46 3.4 Overview of System Hardware 50 3.5 System Software 55 4. OPTIMIZATION OF CELL DETECTION 4.1 C r i t e r i a f o r Optimization 59 4.2 The C e l l Detection Algorithm 60 4.3 Optimization of C e l l Shape 61 4.4 Development of a Focusing Algorithm 4.4.1 Determination of Ideal Focus for a Single C e l l 67 4.4.2 Estimation of the Acceptable Focus Range for C e l l Detection 70 4.4.3 Measurement of Flask Surface Shape 74 4.4.4 A Semi-Automated Focusing Procedure 75 4.4.5 F u l l y Automated Focusing Procedures 80 4.5 S e l e c t i o n of Step Size 84 4.6 Focus Dependence of C e l l Detection Using Selected Step Size....90 5. DEVELOPMENT OF A RECOGNITION ALGORITHM FOR V79 CELLS 5.1 Recognition C r i t e r i a 93 5.2 Data A c q u i s i t i o n 96 5.3 C e l l Signal Features 97 5.4 Feature S e l e c t i o n and Discriminant Function Analysis 102 v i 5.5 P r e - c l a s s i f i c a t i o n of Objects 5.5.1 Imposition of Bounds on Feature Values I l l 5.5.2 Exclusion of Near Neighbours 116 5.6 E f f e c t of Defocusing on C e l l Recognition 5.6.1 Focus E f f e c t s on C e l l Signal Features 118 5.6.2 Recognition Algorithm Performance 120 5.7 P r a c t i c a l Considerations 124 6. MEASUREMENT OF CELL GROWTH AND COLONY SIZE 6.1 Preliminary Considerations 126 6.2 Morphological Features of Growing C e l l Populations 127 6.3 Signal Features of C e l l Colonies 131 6.4 Estimation of Colony Size and Population Density 6.4.1 C o l l e c t i o n of Feature Data 134 6.4.2 C o r r e l a t i o n of Feature Scores with Population Density..134 6.5 Use of V79 Colony Size Information to Dis t i n g u i s h Surviving Colonies from Non-surviving Colonies 6.5.1 Optimization of Factors A f f e c t i n g Sample Size 139 6.5.2 C l a s s i f i c a t i o n of Colonies Using a Single Colony Size Measurement 141 6.5.3 C l a s s i f i c a t i o n of Colonies Using Size and Growth Rate Information 148 6.6 Factors A f f e c t i n g Accuracy of Population Density Estimates 6.6.1 Focus Dependence of Population Density Measurements.... 154 6.6.2 Other Factors A f f e c t i n g Accuracy of Population Density Measurements 159 7. APPLICATION TO LOW DOSE RADIATION STUDIES: MEASUREMENT OF THE RELATIVE BIOLOGICAL EFFECTIVENESS OF DIFFERENT LOW LET RADIATIONS 7.1 Motivation for Measuring the RBE of Low LET Radiations at Low Doses 7.1.1 Influence of LET on B i o l o g i c a l Effectiveness 163 7.1.2 Microdosimetry and Radiation Quality 166 7.1.3 RBE at Low Doses 169 v i i 7.2 Materials and Methods 7.2.1 Radiation Apparatus and Dosimetry 174 7.2.2 C e l l Preparation and I r r a d i a t i o n Procedure 179 7.2.3 P l a t i n g of Samples and Assessment of Su r v i v a l 183 7.3 Results and Analysis 7.3.1 Observations on Results f or Individual Experiments 186 7.3.2 Methods of Analysis 187 7.3.3 RBE Based on a-/3 Parameters 192 7.3.4 RBE Based on Interpolation Between Data Points 199 8. DISCUSSION 8.1 C e l l Analyzer Performance 8.1.1 C e l l Detection and Recognition 202 8.1.2 C l a s s i f i c a t i o n of Colonies 203 8.1.3 Comparison to Previously Used Low Dose Survival Assays 204 8.1.4 Assessment of Errors Associated with the Low Dose Assay 207 8.2 RBE Experiments 8.2.1 Determination of a-/3 and RBE Values 214 8.2.2 Relationship of Results to Models of Radiation Action..217 8.2.3 Sources of Error P a r t i c u l a r to the RBE Experiments 220 8.2.4 Recommendations for Further Investigations 222 9. BIBLIOGRAPHY 224 10. APPENDIX 10.1 Discriminant Function Analysis 234 10.2 Linear-Quadratic Models of Radiation Action 238 v i i i LIST OF TABLES Page Table I Manual c l a s s i f i c a t i o n scheme for V79 c e l l r e cognition algorithm development 98 Table II C e l l s i g n a l features 100 Table III Discriminant function performance for V79 c e l l recognition 110 Table IV E f f e c t of feature bounds on recognition algorithm performance 115 Table V E f f e c t of proximity exclusion on recognition algorithm performance (25 pm exclusion distance) 117 Table VI Performance of the discriminant function method of colony scoring 153 Table VII B i o l o g i c a l effectiveness ( r e l a t i v e to 6 0 C o 7-rays) of selected low LET rad i a t i o n s : d i r e c t observations and predictions based on the Theory of Dual Radiation Action...170 Table VIII C h a r a c t e r i s t i c s of r a d i a t i o n sources used f o r RBE experiments 176 Table IX Dose c a l c u l a t i o n factors for the r a d i a t i o n sources used f o r the RBE experiments 180 Table X Survival curve parameter values for low dose data 197 Table XI Survival curve parameter values for high dose data 197 Table XII RBE l i m i t s f o r D -* 0 and D -> « (D = dose), as ca l c u l a t e d from a and fi parameters 198 Table XIII Survival curve parameter values and RBE l i m i t s f o r combined low dose-high dose f i t s 200 Table XIV RBE's at d i f f e r e n t normalized s u r v i v a l l e v e l s , as obtained by i n t e r p o l a t i o n between data points 201 Table XV Comparison of s u r v i v a l curve parameter values obtained for V79 low dose data with manual scoring of survivors and with scoring using the colony screening method 205 Table XVI Summary of standard deviations f o r the 6 0 C o 7-ray low dose data 210 Table XVII V a r i a t i o n of a-/3 estimates with dose range used f o r f i t ( 6 0Co 7-ray s u r v i v a l data) 216 i x LIST OF FIGURES Page Figure 1 Sampling e r r o r s a s s o c i a t e d w i t h s u r v i v a l experiments 5 Figure 2 Data r e s o l u t i o n i n the f i r s t decade of the c e l l s u r v i v a l curve f o r binomial and Poisson sampling e r r o r s 8 Figure 3 Production of X-rays 13 Figure 4 T y p i c a l mammalian c e l l s u r v i v a l curves f o r exposure to low and high LET r a d i a t i o n s 18 Figure 5 Inverted l i g h t microscopy 31 Figure 6 D i f f r a c t i o n and image r e s o l u t i o n i n an o p t i c a l system 34 Figure 7 Generation of a phase c o n t r a s t image under p a r a l l e l i n c i d e n t i l l u m i n a t i o n 38 Figure 8 Phase c o n t r a s t and modulation c o n t r a s t microscopy 40 Figure 9 V79 c e l l images produced by d i f f e r e n t microscopy modes 43 Figure 10 Image d e t e c t i o n by a l i n e a r CCD sensor under b r i g h t f i e l d and phase c o n t r a s t microscopy modes 44 Figure 11 Image formation i n the modified b r i g h t f i e l d microscopy mode 47 Figure 12 V79 c e l l images produced by the modified b r i g h t f i e l d microscopy mode 48 Figure 13 Image d e t e c t i o n by a l i n e a r CCD sensor under the modified b r i g h t f i e l d microscopy mode 49 Figure 14 The C e l l Analyzer system 51 Figure 15 Block diagram of the C e l l Analyzer 52 Figure 16 Scanning procedure f o r c e l l d e t e c t i o n 54 Figure 17 V79 and CHO c e l l s at d i f f e r e n t times a f t e r t r y p s i n i z a t i o n and p l a t i n g 62 Figure 18 E f f e c t of i n c u b a t i o n time at 37 °C on the shape of f r e s h l y t r y p s i n i z e d and p l a t e d c e l l s 64 Figure 19 Shape changes i n f r e s h l y p l a t e d V79 c e l l s kept a t room temperature a f t e r an i n i t i a l 2 hour p e r i o d a t 37 °C to al l o w f o r c e l l attachment to the p l a t i n g surface 66 Figure 20 E f f e c t of time spent at room temperature (22-25 °C) on c e l l s u r v i v a l 68 Figure 21 E f f e c t of defocusing on c e l l s i g n a l peak height f o r 4.OX o b j e c t i v e s of 3 d i f f e r e n t numerical apertures 71 X Figure 22 E f f e c t of defocusing on c e l l s i g n a l peak height r e l a t i v e to background l i g h t l e v e l s (4.0/0.20 o b j e c t i v e ) 73 Figure 23 The shape of the p l a t i n g surface i n a t y p i c a l Nunclon 25 cm 2 t i s s u e c u l t u r e f l a s k 76 Figure 24 Mean i d e a l focus l e v e l s f o r the 11 scanning bands i n a t y p i c a l Nunclon 25 cm 2 f l a s k 78 Figure 25 Flow chart of the semi-automated f o c u s i n g r o u t i n e f o r Nunclon f l a s k s 79 Figure 26 E f f e c t of defocusing on the c e l l s i g n a l 82 Figure 27 E f f e c t of defocusing on c e l l s i g n a l i n t e n s i t y near the c e l l edges 83 Figure 28 V79 and CHO c e l l s i g n a l peak widths at the d e t e c t i o n t h r e s h o l d 86 Figure 29 Dependence of the number of " h i t s " per V79 c e l l on the step s i z e between successive l i n e scans i n the c e l l d e t e c t i o n procedure 88 Figure 30 E f f e c t of defocusing on V79 c e l l d e t e c t i o n u s i n g a 4 pm step s i z e 91 Figure 31 Confusion matrix and formulae f o r the e v a l u a t i o n of r e c o g n i t i o n a l g o r i t h m performance 94 Figure 32 Geometrical s i g n a l features used i n the c e l l r e c o g n i t i o n a l g o r i t h m 101 Figure 33 D i s t r i b u t i o n s of c e l l and n o n - c e l l feature values 103 Figure 34 D i s t r i b u t i o n of d i s c r i m i n a n t scores i n the l e a r n i n g set used to generate the l i n e a r d i s c r i m i n a n t f u n c t i o n f o r c e l l r e c o g n i t i o n 108 Figure 35 E f f e c t of a l t e r i n g the p o s i t i o n of the d e c i s i o n boundary on d i s c r i m i n a n t f u n c t i o n performance f o r the l e a r n i n g set..109 Figure 36 Use of upper and lower bounds on feature values to i d e n t i f y some n o n - c e l l objects p r i o r to a p p l i c a t i o n of the l i n e a r d i s c r i m i n a n t f u n c t i o n 113 Figure 37 E f f e c t of defocusing on s e l e c t e d c e l l s i g n a l f e a t u r e s 119 Figure 38 E f f e c t of defocusing on r e c o g n i t i o n a l g o r i t h m performance..122 Figure 39 V79 and CHO c o l o n i e s at 3 days and 5 days a f t e r p l a t i n g . . . . 128 Figure 40 Some e f f e c t s of environmental c o n d i t i o n s and damaging agents on c e l l morphology 130 x i Figure 41 Line scans across V79 and CHO c e l l c o l o n i e s 132 Figure 42 S i x of the SSCAN colony features 133 Figure 43 Measurement of V79 p o p u l a t i o n d e n s i t y u s i n g colony f e a t u r e scores ' 136 Figure 44 Measurements of CHO p o p u l a t i o n d e n s i t y by counting peaks i n l i n e scans spaced 6 fim apart 138 Figure 45 E f f e c t of p r o x i m i t y e x c l u s i o n on s i z e of data s et 142 Figure 46 Day 4 " t o t a l area" feature scores f o r s u r v i v i n g and non-s u r v i v i n g V79 c o l o n i e s a r i s i n g from an i r r a d i a t e d (0-2.4 Gy of 250 kV p X-rays) parent c e l l p o p u l a t i o n 144 Figure 47 Day 4 " t o t a l area" feature scores f o r V79 c o l o n i e s a r i s i n g from parent c e l l s i r r a d i a t e d at d i f f e r e n t doses....147 Figure 48 Separation of s u r v i v o r s from non-survivors u s i n g a quadratic d i s c r i m i n a n t f u n c t i o n 151 Figure 49 Focus dependence of some colony features 156 Figure 50 The r a t i o , "volume above"/"area above" (va/aa) as a f u n c t i o n of focus s e t t i n g 160 Figure 51 E f f e c t of a 5-hour p e r i o d at room temperature on colony feature scores 162 Figure 52 C h a r a c t e r i s t i c shape of the RBE versus LET curve f o r mammalian c e l l s 164 Figure 53 Diagram of the experimental set-up used f o r measuring the RBE of low LET r a d i a t i o n s 175 Figure 54 C a l c u l a t e d energy spe c t r a f o r the 55 kV and 250 kV p X-ray beams 177 Figure 55 D i s t r i b u t i o n of c e l l s i n the c e l l c y c l e 184 Figure 56 E s t i m a t i o n of experimental u n c e r t a i n t i e s a s s o c i a t e d w i t h the low and high dose assays 191 Figure 57 Low dose s u r v i v a l data f o r V79 c e l l s 193 Figure 58 Low dose s u r v i v a l data f o r CHO c e l l s 194 Figure 59 High dose s u r v i v a l data f o r V79 c e l l s 195 Figure 60 High dose s u r v i v a l data f o r CHO c e l l s 196 Figure 61 D i v i s i o n of a 2-dimensional feature space by a d i s c r i m i n a n t f u n c t i o n 236 x i i LIST OF ABBREVIATIONS CCD - charge-coupled device. DNA - deoxyribonucleic a c i d , the genetic m a t e r i a l i n c e l l s . Gy - gray, or j o u l e s / k i l o g r a m , a u n i t of r a d i a t i o n dose. keV, MeV - k i l o - e l e c t r o n v o l t s and mega-electron v o l t s , r e s p e c t i v e l y (where 1 e l e c t r o n v o l t = 1.602 X 10" 1 9 j o u l e s ) . kVp - peak voltage i n k i l o v o l t s . In t h i s t h e s i s , r e f e r s to the peak voltage a p p l i e d across an X-ray tube. LET - l i n e a r energy t r a n s f e r (energy deposited per u n i t d i s tance t r a v e l l e d by a charged p a r t i c l e as i t moves through a medium). Quoted values f o r a given r a d i a t i o n modality are ge n e r a l l y averages f o r the spectrum of LET's produced. RBE - r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s of d i f f e r e n t r a d i a t i o n m o d a l i t i e s . U s u a l l y expressed as a r a t i o of r a d i a t i o n doses r e q u i r e d to produce the same l e v e l of e f f e c t f o r two s e l e c t e d r a d i a t i o n m o d a l i t i e s . Sv - s i e v e r t s , which measure "equivalent dose" (with respect to human exposure) f o r r a d i a t i o n s of d i f f e r e n t q u a l i t y . TDRA, GTDRA - Theory of Dual R a d i a t i o n A c t i o n , a theory t h a t p r e d i c t s b i o l o g i c a l response to r a d i a t i o n on the b a s i s of i n t e r a c t i o n of sublesions to produce l e t h a l l e s i o n s . The GTDRA (Generalized Theory of Dual R a d i a t i o n A c t i o n ) i s a m o d i f i c a t i o n of the o r i g i n a l TDRA. x i i i ACKNOWLEDGEMENTS This t h e s i s could not have been completed without the generous c o n t r i b u t i o n s of many people. I would e s p e c i a l l y l i k e to thank my t h e s i s s u p e r v i s o r , Dr. Branko P a l c i c , f o r h i s advice, encouragement, and patience during the course of t h i s work. His f i n a n c i a l support, along w i t h t h a t of the N a t u r a l Sciences and Engineering Research C o u n c i l of Canada, was a l s o of great a s s i s t a n c e to me. I would a l s o l i k e to thank the s t a f f of the Medical B i o p h y s i c s U n i t and the Cancer Imaging S e c t i o n of the B.C. Cancer Research Centre f o r t h e i r t e c h n i c a l a s s i s t a n c e , and f o r making the Research Centre a pleasant place to work. S p e c i a l a p p r e c i a t i o n goes to Steven Poon (Cancer Imaging) f o r i n c o r p o r a t i n g many of my suggestions i n t o the CELREC and CSCAN programs. I am a l s o indebted to the Physics D i v i s i o n , Cancer C o n t r o l Agency of B r i t i s h Columbia, f o r a l l o w i n g me the use of t h e i r radiotherapy u n i t s . I would s p e c i f i c a l l y l i k e to acknowledge the a s s i s t a n c e of W i l l i a m Kwa, who provided me w i t h the beam c h a r a c t e r i s t i c s of the radiotherapy u n i t s , and who helped w i t h a l l aspects of the beam dosimetry and sample i r r a d i a t i o n s f o r the RBE experiments. Without h i s a s s i s t a n c e , these experiments c o u l d not have been performed. F i n a l l y , I would l i k e to thank John Fengler, f o r p r o v i d i n g an e x t r a hand whenever I needed one, and my parents, f o r t h e i r u n f a i l i n g support throughout the course of my s t u d i e s . 1 1. INTRODUCTION 1.1 Measurement of the E f f e c t s of Damaging Agents on C e l l s 1.1.1 Reasons f o r Measuring Damage to Individual C e l l s Largely because of the technological and medical advances made i n the l a s t century, humans are exposed d a i l y to a v a r i e t y of to x i c agents, both chemical and p h y s i c a l i n nature. Exposure to these agents can r e s u l t from t h e i r presence i n the environment, i n consumer products, and i n medical procedures. I t i s therefore imperative to investigate the e f f e c t s of such agents on l i v i n g organisms, both for the assessment of the p o t e n t i a l l y harmful consequences of exposure, and, i n some instances, f or the optimization of therapeutic effectiveness. While damaging agents ultimately i n i t i a t e tissue or whole body responses, there i s often a strong c e l l u l a r basis f or many of the e f f e c t s seen a f t e r exposure. This kind of r e l a t i o n s h i p has been f i r m l y established for many e f f e c t s caused by such extensively studied damaging agents as i o n i z i n g r a d i a t i o n (Alper, 1979). Much can be learned about the mechanism of a c t i o n of damaging agents by studying t h e i r e f f e c t on i n d i v i d u a l c e l l s . Quantitative r e s u l t s are more e a s i l y obtained for large populations i f sing l e c e l l s , rather than m u l t i c e l l u l a r tissues or organisms, are studied. Experiments performed on c e l l s grown i n tissue culture allow environmental and exposure conditions, which may influence the observed e f f e c t , to be characterized and c o n t r o l l e d . A further advantage of such methods i s t h e i r moderate expense i n comparison to animal studies. However, care must always be taken i n r e l a t i n g r e s u l t s obtained i n ti s s u e culture to the more complex m i l i e u that e x i s t s i n a whole animal. 2 1.1.2 Endpoints Used to Quant i fy C e l l u l a r Damage A number of endpoints may be used to study the i n t e r a c t i o n of damaging agents w i t h c e l l s . Included among these are short-term v i a b i l i t y t e s t s t h a t measure the a b i l i t y of c e l l s to take up or exclude c e r t a i n dyes during or soon a f t e r treatment. General morphological or b e h a v i o r a l parameters are a l s o used as endpoints, as are s p e c i f i c genetic or nuclear e f f e c t s . Genetic and nuclear endpoints i n c l u d e the i n d u c t i o n of s p e c i f i c mutations or chromosome a b e r r a t i o n s , the formation of m i c r o n u c l e i , and the p r o d u c t i o n of DNA base damage, s t r u c t u r a l changes, and s t r a n d breaks. Morphological or b e h a v i o r a l changes are measured through such endpoints as c e l l t r a n s f o r m a t i o n , d i v i s i o n delay, and l o s s of p r o l i f e r a t i v e c a p a c i t y . Loss of p r o l i f e r a t i v e c a p a c i t y i n a normally p r o l i f e r a t i n g c e l l p o p u l a t i o n i s f r e q u e n t l y equated w i t h c e l l death (even though n o n - p r o l i f e r a t i n g c e l l s are o f t e n s t i l l m e t a b o l i c a l l y a c t i v e ) , so experiments measuring t h i s endpoint are commonly r e f e r r e d to as s u r v i v a l experiments. The endpoints used i n a p a r t i c u l a r a p p l i c a t i o n are g e n e r a l l y s e l e c t e d according to t h e i r r e l a t i o n s h i p to the phenomenon being i n v e s t i g a t e d . For i n s t a n c e , genetic endpoints, formation of m i c r o n u c l e i , and c e l l t r a n s f o r m a t i o n are a l l of p a r t i c u l a r relevance to the mechanisms of cancer i n d u c t i o n , whereas c e l l s u r v i v a l i s of importance to s t u d i e s of t i s s u e damage, whole body l e t h a l i t y , and e f f i c a c y of cancer treatment p r o t o c o l s . In some cases, endpoints that can be measured are r e s t r i c t e d by the p r o p e r t i e s of the c e l l s being used. S u r v i v a l , f o r example, i s measured f o r i n d i v i d u a l c e l l s by p l a t i n g them i n t o t i s s u e c u l t u r e dishes at s u f f i c i e n t l y low d e n s i t i e s t h a t each s u r v i v i n g c e l l has room to p r o l i f e r a t e i n t o a d i s t i n c t colony, or clone. The c e l l s used f o r such experiments must be capable of a t t a c h i n g to a growth surface and p r o l i f e r a t i n g under these c o n d i t i o n s . Such i s not always the case. F r e s h l y i s o l a t e d , normal 3 mammalian c e l l s , f o r example, are d i f f i c u l t to grow at low dens i t i e s (Freshney, 1987) . Indeed, the f i r s t successful attempt at measuring the s u r v i v a l of i n d i v i d u a l mammalian c e l l s exposed to a l e t h a l agent ( i n th i s case, i o n i z i n g radiation) was not made u n t i l the mid-1950's, when Puck and Marcus developed the necessary techniques to grow sparsely p l a t e d mammalian c e l l s i n tiss u e culture (Puck and Marcus, 1955, 1956; Puck et a l , 1956). Before that time, s u r v i v a l measurements could only be made with b a c t e r i a and f r e e - l i v i n g eukaryotic organisms such as yeast. 1.1.3 Data Requirements and Techniques Employed i n C e l l S u r v i v a l Studies I f quantitative data on the action of damaging agents on l i v i n g systems are to be obtained, care must be taken to ensure that the population sampled i s s u f f i c i e n t l y large to produce s t a t i s t i c a l l y s i g n i f i c a n t r e s u l t s . In the case of endpoints that measure the occurrence r e l a t i v e to the non-occurrence of a p a r t i c u l a r e f f e c t , the experimental outcome w i l l be governed by the binomial p r o b a b i l i t y d i s t r i b u t i o n . This s i t u a t i o n applies to most of the endpoints l i s t e d i n Section 1.1.2, but w i l l here be further described s p e c i f i c a l l y f o r c e l l s u r v i v a l , which i s the endpoint used i n t h i s t h e s i s . In a s u r v i v a l experiment, each c e l l exposed to a p a r t i c u l a r dose of the damaging agent has a c e r t a i n p r o b a b i l i t y , p, of surviving, and a p r o b a b i l i t y q=l-p of not surviving. For a population of s i z e N exposed to a given dose, the surviving f r a c t i o n , S, and i t s variance, a s 2 , would be: S = p ° s 2 = pq/u. ( i . D 4 Since most c e l l s u r v i v a l experiments involve the generation of a dose-response curve, the allowable error i s governed by the need to resolve neighbouring data points i n such a curve, and to d i s t i n g u i s h between curves produced under d i f f e r e n t treatment conditions. I f i t i s assumed that no a d d i t i o n a l sources of error e x i s t , and i f the 95% confidence l i m i t f o r the estimation of S i s taken to represent the uncertainty, SS, associated with any given data point, the minimum allowable d i f f e r e n c e between two resolvable data points, S and S ,,, would be: r n n+1 | S -S | = SS +SS J.1 1 n n+11 n n+1 SS. = 1.96(o- ).. (1.2) i v S' i x ' A p l o t of the minimum percent difference between data points (equal to (S n-S n + 1)xlOO/S n) versus S^, where the higher s u r v i v i n g f r a c t i o n i s assumed to occur at S^, i s shown i n Figure l a . Curves are p l o t t e d f or several values of N, the number of c e l l s assayed. Resolution improves s i g n i f i c a n t l y with increasing N. In p r a c t i c e , however, the a b i l i t y to measure s u r v i v a l f or large numbers of i n d i v i d u a l c e l l s has been severely l i m i t e d , since each treated c e l l must be i d e n t i f i e d and i s o l a t e d before the e f f e c t s of the treatment become apparent. Once i d e n t i f i e d , these c e l l s must be r e v i s i t e d a f t e r an appropriate time i n t e r v a l to assess t h e i r s u r v i v a l . Further contributing to the number of c e l l s that must be assayed i s the f a c t that the generation of a dose-response curve requires measurements to be made at several dose points. At l e a s t two such curves must be generated i n a t y p i c a l experiment. Because of the large number of c e l l s that must be assayed i n s u r v i v a l experiments, dose-response curves have t r a d i t i o n a l l y been generated using methods that only approximate N. These methods involve the determination 5 100 90 80 70 60 50 40 30 20 10 « 0 o £ ioo CO fl • I—< o a fl \V « W W - N = 100 \ \ \ \ — N = 200 S' V N •-N= 500 — N = I O O O N = 2000 ^ io"2 io"1 s u r v i v i n g f r a c t i o n Figure 1. Sampling errors associated with s u r v i v a l experiments. The minimum percent difference between neighbouring data points, (S n-S n + 1)xl00/S n, such that the points S n and S n + 1 are separated by 95% confidence i n t e r v a l s , i s p l o t t e d as a function of the sur v i v i n g f r a c t i o n , S . (a) minimum separation f o r binomial sampling error. (b) minimum separation f o r Poisson sampling error. 6 of c e l l concentrations i n an aqueous suspension, followed by the p l a t i n g of known volumes of suspension into p e t r i dishes. This procedure can be c a r r i e d out with r e l a t i v e ease using a c e l l counter, such as the Coulter Counter (Coulter E l e c t r o n i c s , Hialeah, F l o r i d a ) . These devices determine the number of c e l l s i n a s p e c i f i e d volume of e l e c t r o l y t i c f l u i d by measuring changes i n e l e c t r i c a l resistance as c e l l s i n the suspension are drawn through a small o r i f i c e . The s t a t i s t i c a l errors associated with the above method follow the Poisson p r o b a b i l i t y d i s t r i b u t i o n , since the c e l l s p l ated into a given p e t r i dish are samples of known s i z e taken from a parent population randomly d i s t r i b u t e d within a volume. I t has been shown by Boag (1975) that, i f the binomial response of the c e l l s to the treatment i s considered i n combination with the Poisson variance i n the sample s i z e f or a given s u r v i v a l p r o b a b i l i t y p, the r e s u l t i n g estimate of the s u r v i v i n g f r a c t i o n i s Poisson d i s t r i b u t e d : S - P os* = p/N. (1.3) A p l o t of (S n-S n + 1)xl00/S n versus using Equations 1.2 and 1.3 i s shown i n Figure l b . Comparison with Figure l a indicates that Equations (1.1) and (1.3) y i e l d s i m i l a r r e s o l u t i o n at low s u r v i v a l l e v e l s (as would be expected since the Poisson d i s t r i b u t i o n tends to the binomial for small p and large N) . At surviving f r a c t i o n s above approximately 0.2, however, the r e s o l u t i o n i s s i g n i f i c a n t l y lower for Equation 1.3 than f o r Equation 1.1. Nevertheless, i t i s not d i f f i c u l t to assay 1000 or more c e l l s per dose point using the Poisson sampling method. Indeed, the procedure followed i n most experiments i s to plate increasing numbers of c e l l s as the s u r v i v a l 7 l e v e l decreases, so that the number of s u r v i v i n g colonies per dose point remains approximately constant. This y i e l d s a constant minimum allowable percent d i f f e r e n c e between successive data points: (S -.S JxlOO/S = 200K/(1+*C) n n+1 ' n ' K = l.96(N')~1/z (1.4) where N' i s the number of surviving colonies (N'=pN). The use of t h i s technique s i g n i f i c a n t l y improves r e s o l u t i o n at low s u r v i v a l l e v e l s ( i . e . S < 0.1), allowing dose-response curves to be determined over several decades of c e l l k i l l . This method has proven very u s e f u l i n the i n v e s t i g a t i o n of damaging agents such as i o n i z i n g r a d i a t i o n , which k i l l s c e l l s exponentially with dose. While the conventional Poisson sampling method has been used extensively i n the determination of dose-response r e l a t i o n s h i p s over large dose ranges, i t does not provide adequate r e s o l u t i o n f or d e t a i l e d measurements i n small dose ranges. Figure 2 shows the r e s o l u t i o n of the Poisson sampling method (with N' survivors measured per sample) for 3 s u r v i v a l ranges i n the f i r s t decade of c e l l k i l l . The r e s o l u t i o n that could be achieved i f both s u r v i v a l and non-survival could be measured for a population of known size N (binomial s t a t i s t i c s ) i s also shown for comparison. From the curves i t i s evident that, even with the improvements afforded by keeping A/', rather than N, constant between dose points, the Poisson sampling r e s o l u t i o n i s s t i l l severely l i m i t e d at high s u r v i v a l l e v e l s (S > 0.3). Furthermore, i n p r a c t i c e , a d d i t i o n a l sources of error i n determining N (such as p i p e t t i n g error and inadequate mixing of the suspension p r i o r to counting and/or p l a t i n g c e l l s ) further reduce the r e s o l u t i o n of t h i s method (Bedford and Griggs, 1975). A binomial 8 *-< 25 CO > -g 20-1 •= 15-1 CO •S io-i o a. a* 5 **» 0 0.0 binomial o.k<s«o.9 Poisson 0.5 1.0 1.5 2.0 N or N' ( x 103) Figure 2. Data r e s o l u t i o n i n the f i r s t decade o f the c e l l s u r v i v a l c u r v e f o r b i n o m i a l and P o i s s o n s a m p l i n g e r r o r s ( w i t h 95% c o n f i d e n c e i n t e r v a l s s e p a r a t i n g n e i g h b o u r i n g p o i n t s ) . The maximum number o f d a t a p o i n t s i n v a r i o u s s u r v i v a l i n t e r v a l s i s shown. The r e s o l u t i o n as a f u n c t i o n o f the t o t a l number o f c e l l s a s s a y e d (N) i n a b i n o m i a l e x p e r i m e n t i s compared t o the r e s o l u t i o n as a f u n c t i o n o f the t o t a l number o f s u r v i v i n g c e l l s (W ) f o r a P o i s s o n e x p e r i m e n t , as N and N' a r e , r e s p e c t i v e l y , t he l i m i t i n g q u a n t i t i e s f o r t h e s e two t y p e s o f e x p e r i m e n t s . 9 experiment would a l s o be subject to c e r t a i n procedural e r r o r s , but they would l i k e l y be small given the requirement f o r c a r e f u l s e l e c t i o n and counting of the sample popu l a t i o n . Despite the d i f f i c u l t i e s a s s o c i a t e d w i t h a c c u r a t e l y determining N, some attempts have been made to measure dose-response i n the f i r s t decade of the s u r v i v a l curve. Methods employed have i n c l u d e d i s o l a t i o n and p l a t i n g of i n d i v i d u a l c e l l s w i t h the a i d of a m i c r o p i p e t t e , as w e l l as determination of c e l l l o c a t i o n s r e l a t i v e to some f i x e d coordinate system a f t e r they have been p l a t e d (Bedford and Griggs, 1975; Brosin g , 1983; P a l c i c e t a l , 1984). However, because of the degree of manual e f f o r t r e q u i r e d , only l i m i t e d amounts of data (50-100 c e l l s per data p o i n t per experiment) have been obtained. I d e a l l y , data sets of 400-500 c e l l s per dose p o i n t (or more) are r e q u i r e d to re s o l v e at l e a s t 6 p o i n t s f o r s u r v i v i n g f r a c t i o n s greater than -0.3 ( c f . Figure 2). Some e f f o r t s have been made to achieve samples of t h i s s i z e through automation of the c e l l s e l e c t i o n procedure. As w i t h the manual methods, attempts have been made to s e l e c t c e l l s a u t o m a t i c a l l y e i t h e r before or a f t e r p l a t i n g . These approaches i n c l u d e the counting out of i n d i v i d u a l c e l l s w h i l e they are i n suspension u s i n g a c e l l s o r t e r (Skarsgard et a l , 1987), and the use of a semi-automated microscope-based system to a i d i n the determination of c e l l l o c a t i o n s i n a t i s s u e c u l t u r e f l a s k ( P a l c i c et a l , 1983). The l a t t e r system, however, s t i l l r e q u i r e s considerable manual input i n t o the c e l l s e l e c t i o n procedure. 10 1.1.4 Applications of Image Cytometry to C e l l S u r v i v a l Studies Over the l a s t decade or so, considerable advances have been made i n the development and implementation of image cytometry devices f o r q u a n t i t a t i v e c e l l u l a r s t u d i e s . These devices t y p i c a l l y c o n s i s t of a microscope, an image det e c t o r , an image processor, a computer-controlled microscope stage, and a host computer. Th e i r a p p l i c a t i o n to l i v e c e l l work has, however, been l a r g e l y confined to the generation of time-lapse movies (e.g. Heye et a l , 1982; Tolmach et a l , 1978; Kallman, 1984). Such s t u d i e s are l i m i t e d i n the number of samples that can be analyzed per experiment, and t h e r e f o r e provide p r i m a r i l y q u a l i t a t i v e data. Image cytometry systems are i d e a l l y equipped to c a r r y out f u n c t i o n s such as a u t o m a t i c a l l y scanning a s l i d e or t i s s u e c u l t u r e v e s s e l and r e c o r d i n g the l o c a t i o n s of c e l l s on i t s surface. The automation of these tasks c o u l d g r e a t l y enhance the f e a s i b i l i t y of s u r v i v a l s t u d i e s that r e q u i r e b i n o m i a l sampling. However, while much work has been done i n the development of devices to a u t o m a t i c a l l y l o c a t e s t a i n e d c e l l s viewed on a microscope s l i d e (e.g. Tucker, 1979; Ploem et a l , 1979; Tanaka et a l , 1977) , a s i m i l a r device f o r l i v e c e l l s t u d i e s has only r e c e n t l y been developed ( P a l c i c and J a g g i , 1986). Such a device may a l l o w q u a n t i t a t i v e data to be generated w i t h s u f f i c i e n t speed and p r e c i s i o n to o b t a i n reasonable data r e s o l u t i o n i n the f i r s t decade of the c e l l s u r v i v a l curve. Moreover, because c e l l l o c a t i o n s i n the t i s s u e c u l t u r e f l a s k would be known, other endpoints could be measured c o n c u r r e n t l y w i t h c e l l s u r v i v a l , p e r m i t t i n g c o r r e l a t i o n s between d i f f e r e n t endpoints to be determined on a c e l l - b y - c e l l b a s i s . 11 1.2 Ionizing Radiation as a Damaging Agent 1.2.1 Properties of Ionizing Radiation and D e f i n i t i o n of Radiation Dose Ionizing radiations, which are of s i g n i f i c a n t i n t e r e s t as damaging agents to c e l l s and organisms, are distinguished from other forms of r a d i a t i o n by t h e i r a b i l i t y to ionize molecules i n the medium i n which they are absorbed. They occur i n both p a r t i c u l a t e and electromagnetic forms. P a r t i c u l a t e radiations include electrons, protons, neutrons, and charged ions. While some of these p a r t i c l e s may a r i s e from radioactive decay processes, many of them are used f o r s c i e n t i f i c and/or radiotherapeutic purposes i n the form of high energy beams generated by p a r t i c l e a c c elerators. Electromagnetic radiations e x i s t through a wide spectrum of energies. Although photons with energies on the order of 10 eV are capable of e j e c t i n g electrons from atoms, electromagnetic radiations are c l a s s i f i e d as " i o n i z i n g " at energies greater than 124 eV (which corresponds to wavelengths of less than 10 nm) . Such radiations are i d e n t i f i e d i n the electromagnetic spectrum as X-rays and 7-rays, and can produce ejected electrons with s u f f i c i e n t k i n e t i c energy to cause a d d i t i o n a l i o n i z a t i o n s i n the absorbing matter. Gamma rays are produced by the decay of radioactive isotopes, so t h e i r energy i s c h a r a c t e r i s t i c of the p a r t i c u l a r isotope used. X-rays, on the other hand, are produced when a target (usually tungsten) i s bombarded with high speed electrons accelerated i n a cathode ray tube or p a r t i c l e accelerator. Since the energy of the bombarding electrons i s r a r e l y converted e n t i r e l y into photon energy, the r e s u l t i n g X-rays span a spectrum of energies, ranging from zero to a maximum value corresponding to the peak e l e c t r i c a l p o t e n t i a l through which the bombarding electrons have 12 been a c c e l e r a t e d . Thus, f o r example, the most en e r g e t i c photons produced by a 100 kV p X-ray tube w i l l have energies of 100 keV. In p r a c t i c e , low-energy photons are e l i m i n a t e d from an X-ray beam by passing i t through a f i l t e r . Some of these photons are a l s o f i l t e r e d out by the w a l l of the X-ray tube and by the t a r g e t i t s e l f . Figure 3a shows an example of f i l t e r e d and u n f i l t e r e d X-ray sp e c t r a . X-ray s p e c t r a c o n s i s t of a continuous, or "white", component, but may a l s o e x h i b i t d i s c r e t e , c h a r a c t e r i s t i c l i n e s caused by t r a n s i t i o n s from outer to inner e l e c t r o n s h e l l s i n the t a r g e t atoms. T r a n s i t i o n s occur a f t e r an e l e c t r o n from an inner s h e l l has been e j e c t e d from the atom by an incoming e l e c t r o n (Figure 3b). The continuous spectrum, on the other hand, i s produced when the bombarding e l e c t r o n i s decele r a t e d or, o c c a s i o n a l l y , completely stopped by a t a r g e t nucleus. The energy l o s t i n these i n t e r a c t i o n s appears as the continuous, or bremsstrahlung, r a d i a t i o n . R a d i a t i o n i n t e r a c t s w i t h matter through energy d e p o s i t i o n i n the form of i o n i z a t i o n s and e x c i t a t i o n s . I o n i z i n g r a d i a t i o n dose i s there f o r e g e n e r a l l y expressed i n terms of the energy deposited per u n i t mass. In SI u n i t s , t h i s i s given i n j o u l e s per kilogram, or gray (Gy). Energy d e p o s i t i o n by i o n i z i n g r a d i a t i o n can r e s u l t i n the d i s r u p t i o n of molecular bonds i n the absorbing m a t e r i a l . For electromagnetic r a d i a t i o n s , most of t h i s damage does not occur d i r e c t l y , but through the production of fast-moving secondary e l e c t r o n s . These e l e c t r o n s are generated e i t h e r through the p h o t o e l e c t r i c e f f e c t , p a i r production, or, most p r e v a l e n t l y f o r r a d i o b i o l o g i c a l or r a d i o t h e r a p e u t i c a p p l i c a t i o n s , Compton s c a t t e r i n g (see, f o r example, Johns and Cunningham, 1983 or Young, 1983). The t o t a l dose deposited by a r a d i a t i o n i s not the only f a c t o r that determines i t s b i o l o g i c a l e f f e c t i v e n e s s . The p a t t e r n of energy d e p o s i t i o n , or r a d i a t i o n q u a l i t y , i s a l s o important. This has l e d to the i n t r o d u c t i o n 1 3 (a) cn o o o G > • I - H CO OJ \ unf i l tered \ — f i l t e r e d V character is t ic V r a d i a t i o n \ / maximum J energy of incident x^electrons p h o t o n e n e r g y (b) a t o m inc ident e lectron, character is t ic r a d i a t i o n decelerated electron i n c i d e n t electron incident electron b r e m s s t r a h l u n g r a d i a t i o n a tomic nucleus decelerated electron ejected electron Figure 3. Production of X-rays. (a) spectrum of photon energies produced when electrons bombard a target i n an X-ray tube, both before a n d a f t e r f i l t r a t i o n . (b) mechanisms through which X-rays are produced b y i n t e r a c t i o n s between the incident electrons and the target a t o m s . C h a r a c t e r i s t i c r a d i a t i o n r e s u l t s from t r a n s i t i o n s between outer and inner e l e c t r o n s h e l l s i n the target atom a f t e r an incident e l e c t r o n has e j e c t e d an atomic electron from an inner s h e l l ( i l l u s t r a t e d for the K-sh e l l i n t h e f i g u r e ) . Bremsstrahlung r a d i a t i o n occurs when incident electrons a r e decelerated or stopped upon passing near a target nucleus. 14 of the concept of "dose equivalence", which i s used e s p e c i a l l y i n the f i e l d of r a d i a t i o n p r o t e c t i o n . The dose e q u i v a l e n t , H, i s d e f i n e d as: H = D Q (1.5) where D i s the absorbed dose and Q i s a dimensionless " q u a l i t y f a c t o r " f o r the type of r a d i a t i o n i n question (ICRU, 1986). For dose measured i n gray, the u n i t s of H are s i e v e r t s (Sv). The q u a l i t y f a c t o r , Q, i s r e l a t e d to the r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s (RBE) of the r a d i a t i o n of i n t e r e s t i n r e l a t i o n to a standard reference r a d i a t i o n ( u s u a l l y X-rays or 7-rays) under a given s et of exposure c o n d i t i o n s . At a c e r t a i n l e v e l of e f f e c t , p: RBE(p) = D r(p)/D(p) (1.6) where D(p) i s the absorbed dose of the r a d i a t i o n of i n t e r e s t , and D r(p) i s the absorbed dose of the reference r a d i a t i o n . Because of the manner i n which i t i s defined, the RBE f o r a given type of r a d i a t i o n i s dependent on the type and l e v e l of e f f e c t being measured. Q u a l i t y f a c t o r s used i n r a d i a t i o n p r o t e c t i o n have t h e r e f o r e of n e c e s s i t y been d e r i v e d from b i o l o g i c a l data s e l e c t e d on a somewhat s u b j e c t i v e b a s i s . C e r t a i n p h y s i c a l parameters can be used to describe the p a t t e r n of energy d e p o s i t i o n by i o n i z i n g r a d i a t i o n and can, t h e r e f o r e , be r e l a t e d to Q. One such parameter i s the l i n e a r energy t r a n s f e r , or LET. This q u a n t i t y , which can be c a l c u l a t e d t h e o r e t i c a l l y , has become w e l l e s t a b l i s h e d as a means of s p e c i f y i n g r a d i a t i o n q u a l i t y , both i n r a d i a t i o n p r o t e c t i o n and i n r a d i o b i o l o g y . By d e f i n i t i o n (ICRU, 1970a, 1980), LET i s a measure of the energy deposited per u n i t d i s tance t r a v e l l e d by a charged p a r t i c l e as i t moves through a medium: 15 LET A = (d£/di) A (1.7) where E i s energy, i i s distance along the p a r t i c l e t r a c k , and A s p e c i f i e s an energy c u t o f f value a p p l i e d to secondary e l e c t r o n s from p a r t i c u l a r l y e n e r g e t i c i n t e r a c t i o n s i n the primary t r a c k . The secondary t r a c k s a r i s i n g from such e l e c t r o n s are r e f e r r e d to as d e l t a rays, and are excluded from the LET c a l c u l a t i o n i f the i n i t i a l energy of the secondary e l e c t r o n exceeds A ( i n eV) . Values f o r u n r e s t r i c t e d LET (A = •») are, however, commonly used, and the s u b s c r i p t on LET^ i s o f t e n omitted when such values are quoted. LET values are u s u a l l y given i n keV/^m. The LET f o r a charged p a r t i c l e i s dependent on both i t s v e l o c i t y and charge. In general, low energy heavy charged p a r t i c l e s have a h i g h LET, whereas hi g h energy e l e c t r o n s have a low LET. In the case of electromagnetic r a d i a t i o n s , LET values are c a l c u l a t e d from the t r a c k s of the secondary e l e c t r o n s produced. Since these e l e c t r o n s possess a spectrum of energies even f o r monoenergetic photon beams, c h a r a c t e r i s t i c LET's f o r electromagnetic r a d i a t i o n s must be expressed as averages over the e l e c t r o n s p e c t r a generated. Furthermore, the LET of a given p a r t i c l e t r a c k must a l s o be expressed as an average, s i n c e LET increases as the p a r t i c l e i s slowed down by i t s i n t e r a c t i o n s w i t h the absorbing medium. This average i s most commonly c a l c u l a t e d by d i v i d i n g the t r a c k i n t o equal lengths and f i n d i n g the mean energy deposited per u n i t length. Other q u a n t i t i e s that c h a r a c t e r i z e r a d i a t i o n q u a l i t y have been de f i n e d u s i n g microdosimetric techniques, which were developed e x p r e s s l y f o r t h i s purpose ( R o s s i , 1959). Microdosimetry has t r a d i t i o n a l l y i n v o l v e d the d i r e c t measurement of energy d e p o s i t i o n i n small volumes ( t y p i c a l l y on the order of 1 um i n diameter) us i n g p r o p o r t i o n a l counters made from t i s s u e - e q u i v a l e n t m a t e r i a l . Computational methods u t i l i z i n g Monte Carlo s i m u l a t i o n s of charged p a r t i c l e t r a c k s have now a l s o become an important p a r t of the f i e l d ( K e l l e r e r , 1985). Among the most commonly measured microdosimetric q u a n t i t i e s are the l i n e a l energy, y, and the s p e c i f i c energy, z. By d e f i n i t i o n (ICRU, 1980): where i and m are the mean chord le n g t h and mass, r e s p e c t i v e l y , of the volume of i n t e r e s t , and e i s the energy deposited. These q u a n t i t i e s are a f f e c t e d by v a r i o u s f a c t o r s r e l a t e d to p a t t e r n s of energy d e p o s i t i o n , i n c l u d i n g LET, t r a c k length, number of t r a c k s , and t r a c k curvature (ICRU, 1986) . 1.2.2 A c t i o n of R a d i a t i o n on L i v i n g C e l l s The i o n i z a t i o n s and e x c i t a t i o n s created when r a d i a t i o n i n t e r a c t s w i t h matter can cause damage to l i v i n g c e l l s through two d i f f e r e n t mechanisms. C r i t i c a l macromolecules w i t h i n the c e l l may be damaged d i r e c t l y by c o l l i s i o n s w i t h i o n i z i n g p a r t i c l e s , or they may be damaged by f r e e r a d i c a l s produced by i n t e r a c t i o n s between i o n i z i n g p a r t i c l e s and other molecules i n the c e l l . In the l a t t e r category, r a d i o l y s i s products of water are thought to be the primary damaging agents. The most important r e a c t i o n s are considered to be: 7 = e/£ z = e/m (1.8) (1.9) H„0 -»• H„0+ + e" 2 2 H 20 + + H20 - H 30 + + OH" (1.10) (1.11) where r e a c t i o n (1.10) occurs when a water molecule i s i o n i z e d by r a d i a t i o n , and r e a c t i o n (1.11) r e s u l t s when the s h o r t - l i v e d (~10~10 seconds) product, 17 H 20 +, r e a c t s w i t h a water molecule. R a d i o l y s i s products may r e a c t w i t h each other to form s t a b l e molecules, but i t has been estimated t h a t the OH' r a d i c a l , through i t s i n t e r a c t i o n w i t h macromolecules, i s r e s p o n s i b l e f o r approximately two-thirds of c r i t i c a l X-ray damage to mammalian c e l l s (Roots and Okada, 1975). Once c e l l u l a r damage has occurred through e i t h e r d i r e c t or i n d i r e c t mechanisms, biochemical r e p a i r processes may f u r t h e r c o n t r i b u t e to the e f f e c t s that are u l t i m a t e l y manifested. These processes can f u l l y r e s t o r e damaged c e l l components, but may a l s o " m i s r e p a i r " some l e s i o n s , thus " f i x i n g " the damage. F i x a t i o n of damage can a l s o occur through chemical r e a c t i o n s between damaged macromolecules and molecules such as oxygen, which can combine w i t h damaged s i t e s i n such a way that they are i r r e p a r a b l e ( Q u i n t i l i a n i , 1979). Damage f i x a t i o n through such chemical r e a c t i o n s occurs i n the order of microseconds or l e s s , w h i l e c e l l u l a r r e p a i r processes occur over time spans ranging from a few minutes to s e v e r a l days (Tobias et a l , 1980). Even then, the r e s i d u a l damage may not be f u l l y expressed unless the c e l l attempts to d i v i d e . When c e l l s i n t i s s u e c u l t u r e are i r r a d i a t e d , a c h a r a c t e r i s t i c p a t t e r n of s u r v i v a l as a f u n c t i o n of dose i s observed. When p l o t t e d on a semi-logarithmic graph, mammalian c e l l s u r v i v a l a f t e r exposure to low LET r a d i a t i o n s e x h i b i t s a dose-response c h a r a c t e r i z e d by a "shoulder" s e p a r a t i n g regions of d i f f e r i n g slope. The shoulder can vary c o n s i d e r a b l y i n s i z e f o r d i f f e r e n t c e l l l i n e s , but i s almost always present a f t e r exposure to low LET r a d i a t i o n s ( H a l l , 1988). High LET r a d i a t i o n s , on the other hand, produce a s t r i c t l y e xponential dose response (see Figure 4 ) . Evidence suggests that nuclear DNA i s the c r i t i c a l t a r g e t f o r r a d i a t i o n - i n d u c e d c e l l l e t h a l i t y (Munro, 1970; Grote et a l , 1981b, Marin and Bender, 1963). Nevertheless, the exact nature of the b i o p h y s i c a l 18 dose (Gy) Figure 4. T y p i c a l mammalian c e l l s u r v i v a l c u r v e s f o r e x p o s u r e t o low and h i g h LET r a d i a t i o n s . 19 events r e s p o n s i b l e f o r the shape of the mammalian c e l l s u r v i v a l curve has not been e l u c i d a t e d due to the complexity of the processes i n v o l v e d . Despite these c o m p l e x i t i e s , however, many models and t h e o r i e s have been put forward to account f o r the shape of the s u r v i v a l curve. These models are based on v a r i o u s assumptions regarding both the mechanisms through which damage i s i n c u r r e d , and the nature of the processes l e a d i n g to c e l l death. The r e s u l t i n g mathematical formulations can be used to f i t s u r v i v a l data, and thereby a i d i n the i n t e r p r e t a t i o n of experimental r e s u l t s . In a d d i t i o n , models can be used to p r e d i c t r a d i a t i o n e f f e c t s under v a r i o u s c o n d i t i o n s , and so can be t e s t e d to some extent. However, as p o i n t e d out by s e v e r a l authors (e.g. H a l l , 1988; M i l l a r et a l , 1978; P a l c i c et a l , 1985; A l p e r , 1980), conventional measurements of c e l l s u r v i v a l cannot be made w i t h s u f f i c i e n t accuracy or scope to determine the v a l i d i t y of most models on the b a s i s of curve f i t t i n g alone. Furthermore, the i n a b i l i t y to d i s t i n g u i s h between d i f f e r e n t s u r v i v a l models on t h i s b a s i s becomes more pronounced as the number of a d j u s t a b l e parameters i n c r e a s e s . Indeed, f o r a t y p i c a l s u r v i v a l curve c o n s i s t i n g of 8-12 measured dose p o i n t s , only two parameters can be f i t t e d w i t h reasonable p r e c i s i o n (e.g. w i t h i n ±10%) (Chapman, 1980). 1.2.3 Importance of Studying Low Dose R a d i a t i o n E f f e c t s Since the d i s c o v e r y of X-rays by Roentgen i n 1895, r a d i a t i o n has had a dramatic impact on the development of medical technology. This i n f l u e n c e began w i t h the i n t r o d u c t i o n of the roentgenogram, or "X-ray", as a means of imaging the i n t e r n a l s t r u c t u r e s of the human body, and was soon f o l l o w e d by the o b s e r v a t i o n that exposure to X-rays and other forms of i o n i z i n g r a d i a t i o n c o u l d cause damage to b i o l o g i c a l t i s s u e s . The t h e r a p e u t i c value of t h i s l a t t e r phenomenon was soon recognized, and before long r a d i a t i o n 20 was being used to t r e a t a v a r i e t y of i l l n e s s e s , i n c l u d i n g cancer, which i s the primary t a r g e t of radiotherapy today. In the 1920's, i t was discovered through e m p i r i c a l o b s e r v a t i o n that d e l i v e r y of the treatment dose i n a s e r i e s of f r a c t i o n s was more e f f e c t i v e than a s i n g l e acute dose f o r tumour e r a d i c a t i o n w i t h a minimum of normal t i s s u e damage. E l u c i d a t i o n of the mechanisms u n d e r l y i n g such observations could, however, only be achieved through s c i e n t i f i c i n v e s t i g a t i o n of the e f f e c t s of r a d i a t i o n on c e l l s and t i s s u e s . Much progress has been made i n t h i s area s i n c e the development of techniques f o r the q u a n t i t a t i v e study of r a d i a t i o n e f f e c t s on i n d i v i d u a l mammalian c e l l s . Studies of clonogenic s u r v i v a l , both in vivo and in vitro, have l e d to the c o n c l u s i o n that the f r a c t i o n a t i o n of r a d i a t i o n dose decreases damage to normal t i s s u e s by a l l o w i n g r e p a i r and r e p o p u l a t i o n i n the i n t e r v a l between f r a c t i o n s . While these phenomena a l s o occur i n tumours, r e p a i r c a p a c i t y may be diminished i n these t i s s u e s (Denekamp, 1982). Tumours a l s o possess other p r o p e r t i e s that may le a d to increased damage i f f r a c t i o n a t e d , r a t h e r than acute, r a d i a t i o n doses are d e l i v e r e d . Due to t h e i r r a p i d growth, they may co n t a i n p o o r l y oxygenated (and the r e f o r e r a d i o r e s i s t a n t ) c e l l s i n regions f a r from the nearest blood supply. When a f r a c t i o n a t e d treatment regimen i s employed, p r e v i o u s l y hypoxic c e l l s may be reoxygenated between dose f r a c t i o n s due to c e l l k i l l i n g i n regions of the tumour c l o s e r to the blo o d supply. F r a c t i o n a t i o n would th e r e f o r e a l l o w more e f f e c t i v e treatment of the hypoxic regions of a tumour than would a s i n g l e acute dose. C e l l p r o l i f e r a t i o n , which g e n e r a l l y occurs to a l a r g e r degree i n tumours than i n normal t i s s u e , a l s o a f f e c t s r a d i a t i o n response. In a d d i t i o n to the f a c t that c e l l death as a r e s u l t of r a d i a t i o n damage normally manifests i t s e l f when the c e l l attempts to d i v i d e (Denekamp, 21 1982) , c e l l s i n d i f f e r e n t phases of the c e l l c y c l e show d i f f e r e n t i a l r a d i o s e n s i t i v i t y . R a d i a t i o n treatment a l s o delays c e l l d i v i s i o n i n a dose-dependent f a s h i o n ( E l k i n d et a l , 1963), and may a f f e c t c e l l d i v i s i o n r a t e s ( S i n c l a i r , 1964). A f t e r treatment w i t h a r a d i a t i o n dose, these f a c t o r s may combine to r e d i s t r i b u t e the p r o l i f e r a t i n g c e l l s i n t o d i f f e r e n t phases of the c e l l c y c l e .in such a way t h a t the p r o p o r t i o n of c e l l s i n more s e n s i t i v e phases would be enhanced when the next dose f r a c t i o n was a p p l i e d (Zeman and Bedford, 1984). In a d d i t i o n , quiescent c e l l s may be r e c r u i t e d i n t o a p r o l i f e r a t i v e s t a t e a f t e r a r a d i a t i o n dose (Kallman et a l , 1979), which may increase t h e i r s e n s i t i v i t y to subsequent exposures (but u n f o r t u n a t e l y a l s o causes tumour r e p o p u l a t i o n ) . Thus, the enhanced e f f e c t of f r a c t i o n a t i o n on tumour k i l l r e l a t i v e to normal t i s s u e damage appears to be dependent on r a t h e r complex i n t e r a c t i o n s between c e l l u l a r subpopulations that r e a c t i n d i f f e r e n t ways to r a d i a t i o n damage. Indeed, tumour heterogeneity i s a major c o m p l i c a t i n g f a c t o r i n attempts to develop and assess new treatment s t r a t e g i e s ( S u i t and Walker, 1989). Nevertheless, much of what has been deduced about the u n d e r l y i n g mechanisms th a t govern the radioresponse of these subpopulations has been discovered u s i n g c u l t u r e d c e l l s or animal tumour models. However, even w i t h these r e l a t i v e l y simple systems, measurements of r a d i a t i o n e f f e c t s have t y p i c a l l y been made at doses much higher than those employed i n a t y p i c a l t h e r a p e u t i c dose f r a c t i o n . A s i n g l e treatment i n a f r a c t i o n a t e d treatment regimen normally i n v o l v e s doses i n the 1 to 3 Gy range, producing s u r v i v i n g f r a c t i o n s that are g e n e r a l l y greater than 0.1. S u r v i v a l curves generated in vitro, on the other hand, t y p i c a l l y span at l e a s t 2 or 3 decades of c e l l k i l l , w h i l e in vivo colony s u r v i v a l measurements (e.g. Hewitt and Wilson, 1959; T i l l and McCulloch, 1961; Withers and E l k i n d , 1969) are l i m i t e d to s u r v i v i n g 22 f r a c t i o n s below approximately 0.01 (Zeman and Bedford, 1984). The r e s o l u t i o n of both these methods has been l i m i t e d by the i n a b i l i t y to count the number of t r e a t e d c e l l s w i t h s u f f i c i e n t accuracy. E f f e c t s at t h e r a p e u t i c doses are t h e r e f o r e t y p i c a l l y e x t r a p o l a t e d from h i g h dose data u s i n g s u r v i v a l curve models. However, while many of these equations f i t h i g h dose s u r v i v a l data e q u a l l y w e l l , they can p r e d i c t s i g n i f i c a n t l y d i f f e r e n t responses when e x t r a p o l a t e d to low doses (Goodhead, 1980; P a l c i c et a l , 1985). Thus, the generation of low dose s u r v i v a l data i s important to provide e x p l i c i t i n f o r m a t i o n at t h e r a p e u t i c doses, and may a l s o c o n t r i b u t e to the o v e r a l l understanding of mechanisms of r a d i a t i o n a c t i o n on i n d i v i d u a l c e l l s . 1.3 Thesis Objectives The aims of t h i s t h e s i s were p r i m a r i l y concerned w i t h the i n v e s t i g a t i o n of c e l l u l a r radioresponse i n the f i r s t decade of the c e l l s u r v i v a l curve, w i t h emphasis on the use of image cytometry to f a c i l i t a t e these measurements. The major o b j e c t i v e s were: 1) To t e s t the hypothesis that procedures and algorithms can be developed to d i r e c t an automated image cytometry device to a c c u r a t e l y detect, recognize, and record the l o c a t i o n s of l i v e c e l l s p l a t e d i n t o a t i s s u e c u l t u r e f l a s k , w i t h p a r t i c u l a r a t t e n t i o n to speed, accuracy, and o p t i m i z a t i o n of sample s i z e . 2) To t e s t the hypothesis that such a device could a l s o be used to make c e l l p o p u l a t i o n d e n s i t y estimates, w i t h the primary aim being the 23 implementation of automated methods to count s u r v i v o r s and non-survivors i n i r r a d i a t e d samples. 3) I f (1) and (2) were proven, to d i r e c t l y t e s t the p r e d i c t i o n s of the Theory of Dual R a d i a t i o n A c t i o n ( K e l l e r e r and R o s s i , 1972) as a p p l i e d to low LET r a d i a t i o n s . In p a r t i c u l a r , the r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s of s e l e c t e d low LET r a d i a t i o n s i n the f i r s t decade of the c e l l s u r v i v a l curve was to be measured and compared w i t h the p r e d i c t i o n s of the theory. The r a d i a t i o n m o d a l i t i e s t e s t e d were: 6 0 C o 7-rays, 250 kV p X-rays, 55 kV p X-rays, and 11 MeV e l e c t r o n s . 4) To assess the performance of the procedures developed f o r (1) and (2) on the b a s i s of the s t a t i s t i c a l accuracy of the experimental data. 5) To i n t e r p r e t the experimental r e s u l t s i n r e l a t i o n to models of r a d i a t i o n a c t i o n , and i n terms of t h e i r relevance to c u r r e n t l y used r a d i o t h e r a p e u t i c treatment p r o t o c o l s . 24 2. CELL HANDLING AND PREPARATION 2.1 V 7 9 C e l l C u l t u r e The Chinese hamster V79-171 c e l l l i n e was used f o r most of the work desc r i b e d i n t h i s t h e s i s . These c e l l s were r o u t i n e l y grown as attached monolayers i n 25 cm 2 polystyrene t i s s u e c u l t u r e f l a s k s (Falcon) i n a h u m i d i f i e d incubator at 37 °C. E s s e n t i a l n u t r i e n t s and growth f a c t o r s r e q u i r e d f o r c e l l p r o l i f e r a t i o n were s u p p l i e d by a growth medium c o n s i s t i n g of Eagle's minimum e s s e n t i a l medium (F15 n u t r i e n t mixture) supplemented w i t h 10% f e t a l c a l f serum. In a d d i t i o n to h a n d l i n g the c e l l s under s t e r i l e c o n d i t i o n s at a l l times, a n t i b i o t i c s (70 IU/ml p e n i c i l l i n and 70 /xg/ml streptomycin) were added to the growth medium to i n h i b i t b a c t e r i a l contamination. Sodium bicarbonate (2 g/1) was a l s o added as a b u f f e r to maintain pH i n the 7.2-7.4 range upon exposure to a 5% C0 2, 95% a i r atmosphere, which was present i n the incubator i n which the f l a s k s were sto r e d . Exponential growth of the c e l l p o p u l a t i o n was maintained by d i l u t i o n and t r a n s f e r to f r e s h t i s s u e c u l t u r e f l a s k s 2 to 3 times per week. This s u b c u l t u r i n g procedure r e q u i r e d the c e l l monolayer to be detached from the f l a s k and d i s p e r s e d to form a c e l l suspension. This was achieved by removing the growth medium from the f l a s k and exposing the c e l l s to a small volume (< 1 ml) of t r y p s i n s o l u t i o n (0.1% t r y p s i n i n c i t r a t e b u f f e r ) . T r y p s i n i s a p r o t e o l y t i c enzyme t h a t , upon exposures of short d u r a t i o n , breaks down c e l l - s u r f a c e - a s s o c i a t e d p r o t e i n s r e s p o n s i b l e f o r c e l l - s u b s t r a t e and c e l l - c e l l adhesion. I t s a c t i o n i s i n h i b i t e d by c e r t a i n c o n s t i t u e n t s i n the growth medium. Consequently, the growth medium had to be removed from the f l a s k s p r i o r to t r y p s i n i z a t i o n by pouring i t o f f and r i n s i n g once w i t h 2-3 ml of t r y p s i n s o l u t i o n . A f t e r a 5-6 minute exposure to t r y p s i n at 37 °C, c e l l s would detach from t h e i r growth surface and co u l d be suspended i n f r e s h growth medium (10 ml) . The c o n c e n t r a t i o n of c e l l s i n the r e s u l t i n g suspension was determined u s i n g a C o u l t e r Counter. F i n a l l y , a l i q u o t s c o n t a i n i n g 0.2-1.0 X 10 5 c e l l s were added to new t i s s u e c u l t u r e f l a s k s c o n t a i n i n g 5 ml of f r e s h growth medium. These samples were then pla c e d i n the incubator. 2.2 CHO C e l l C u l t u r e Various measurements were a l s o c a r r i e d out on a second c e l l l i n e , the Chinese hamster ovary (CHO) l i n e . CHO c e l l s were grown i n c u l t u r e i n Eagle's minimum e s s e n t i a l medium (a n u t r i e n t mixture) supplemented w i t h 10% f e t a l c a l f serum, a n t i b i o t i c s , and sodium bicarbonate. These c e l l s could be maintained i n monolayer c u l t u r e i n the same f a s h i o n as V79 c e l l s , or cou l d be grown as a s i n g l e c e l l suspension i n s p e c i a l g l a s s spinner f l a s k s . CHO c e l l s growing i n suspension separate a f t e r m i t o s i s , whereas V79 c e l l s form m u l t i c e l l aggregates that are subject to i n t e r n a l g r a dients of oxygen and other n u t r i e n t s (Sutherland et a l , 1971, Franko and Sutherland, 1979). C e l l s were prevented from s e t t l i n g i n the suspension c u l t u r e s by constant s t i r r i n g at 200 rpm. Spinner c u l t u r e s were d i l u t e d d a i l y w i t h f r e s h growth medium i n order to keep the c e l l c o n c e n t r a t i o n between 0.7 X 10 s and 4.0 X 10 s c e l l s / m l , thus a l l o w i n g the p o p u l a t i o n to remain i n exponential growth phase. Unless otherwise mentioned, CHO c e l l s grown i n suspension c u l t u r e were used. 26 2.3 P r e p a r a t i o n o f F l a s k s f o r the C e l l A n a l y z e r To prepare monolayer c u l t u r e c e l l s f o r measurements w i t h the C e l l A nalyzer, 2 day o l d , 50-80% confluent monolayers were t r y p s i n i z e d , suspended, and counted as described above ( S e c t i o n 2.1). C e l l s maintained i n suspension c u l t u r e c o u l d be removed d i r e c t l y from the spinner f l a s k . For a n a l y s i s , c e l l s were p l a t e d i n t o 25 cm 2 t i s s u e c u l t u r e f l a s k s (Nunclon #152094) c o n t a i n i n g 5 ml of growth medium. Between 2000 and 12000 c e l l s were p l a t e d per f l a s k , y i e l d i n g p l a t i n g d e n s i t i e s of 1-5 cells/mm 2 of f l a s k s u r f a ce. These p l a t i n g d e n s i t i e s provided s u f f i c i e n t space around most c e l l s to a l l o w measurements on i n d i v i d u a l c e l l s to be made. The c e l l s were allowed to s e t t l e and adhere to the bottoms of the f l a s k s by i n c u b a t i n g them at 37 °C f o r approximately 1 hour, a f t e r which they were b r i e f l y removed from the incubator i n order to completely f i l l the f l a s k s w i t h growth medium. This acted to reduce a g i t a t i o n of the medium during h a n d l i n g of the samples, and a l s o prevented the formation of surface bubbles t h a t c o u l d i n t e r f e r e w i t h s i g n a l formation by the C e l l Analyzer. 27 3. THE CELL ANALYZER 3.1 Important Considerations Influencing the System Design The image cytometry system used f o r t h i s t h e s i s was the C e l l A nalyzer. This automated device, described i n d e t a i l by J a g g i e t a l (1986) and P a l c i c and J a g g i (1989), was developed a t the B.C. Cancer Research Centre f o r the d e t e c t i o n and a n a l y s i s of l i v e c e l l s growing i n t i s s u e c u l t u r e . The a p p l i c a t i o n to l i v e c e l l s t u d i e s r e q u i r e d t h a t s e v e r a l important design c o n s i d e r a t i o n s be addressed. Foremost among these was the need to perform measurements as q u i c k l y as p o s s i b l e . L i v e c e l l s are not s t a t i c e n t i t i e s . They may, f o r example, change shape, d i v i d e , and sometimes even move appreciable distances i n the span of a few hours. Furthermore, due to the v a r i a b i l i t y inherent i n any c e l l p o p u l a t i o n , l a r g e numbers of c e l l s must g e n e r a l l y be analyzed i n order to measure p o p u l a t i o n c h a r a c t e r i s t i c s . These c e l l s , t y p i c a l l y p l a t e d at low d e n s i t i e s i n order to minimize i n t e r f e r e n c e between neighbours, must f i r s t be l o c a t e d before f u r t h e r a n a l y s i s can be undertaken. This l a t t e r requirement meant that r e l a t i v e l y l a r g e areas (on the order of 10-20 cm 2 f o r a t y p i c a l t i s s u e c u l t u r e v e s s e l ) had to be scanned i n a short p e r i o d of time. For a t y p i c a l c e l l s u r v i v a l experiment, t h i s scanning time would have to be 10-15 minutes per f l a s k or b e t t e r . Even w i t h sampling at r e l a t i v e l y low s p a t i a l r e s o l u t i o n (e.g. every 3 nm, which i s about one t h i r d of a t y p i c a l c e l l diameter) , t h i s would mean that at l e a s t 10 s p o i n t s would have to be sampled per f l a s k at a data r a t e of 10 megabytes per minute. L i v e c e l l s are e s s e n t i a l l y transparent, making them d i f f i c u l t to detect under standard b r i g h t f i e l d microscopy modes. An o p t i c a l c o n f i g u r a t i o n that e x p l o i t e d the r e f r a c t i v e , r a t h e r than the ab s o r p t i v e , 28 p r o p e r t i e s of the c e l l s t h e r e f o r e had to be u t i l i z e d . I n a d d i t i o n , due to the a p p r e c i a b l e amount of debris commonly present i n a t i s s u e c u l t u r e p r e p a r a t i o n , i t was d e s i r a b l e that the c e l l image have a c h a r a c t e r i s t i c s i g n a l (a "signature") that would a l l o w i t to be e a s i l y d i s t i n g u i s h e d from other o b j e c t s i n the sample. Because of the speed requirements f o r the system, the d i s t i n c t i o n between c e l l s and n o n - c e l l s would have to be made at the lowest p o s s i b l e r e s o l u t i o n . L a s t l y , measurements c a r r i e d out by the C e l l Analyzer had to be made under c o n d i t i o n s that d i d not perturb the c e l l s unduly. Furthermore, i n order to a l l o w comparisons w i t h data i n the l i t e r a t u r e , i t was d e s i r a b l e t h a t the c o n d i t i o n s under which experiments were performed were as cl o s e as p o s s i b l e to those that have t r a d i t i o n a l l y e x i s t e d i n s u r v i v a l s t u d i e s . Thus, samples co u l d not be viewed under c o n d i t i o n s of h i g h l i g h t i n t e n s i t y , frequent and r a p i d a c c e l e r a t i o n s and d e c e l e r a t i o n s , temperature extremes, or a d d i t i o n of p o t e n t i a l l y t o x i c substances (such as s t a i n s ) to t h e i r growth environment. To maintain a s e p t i c c o n d i t i o n s , c e l l s a l s o had to be kept i n s t e r i l e , sealed v e s s e l s constructed of m a t e r i a l s , such as s p e c i a l l y t r e a t e d t i s s u e c u l t u r e p l a s t i c s , that provided a s u i t a b l e growth surface f o r the c e l l s . 3.2 Image Detection Among the most important components of an image cytometry system i s the image sensor. This component i n f l u e n c e s the speed, r e s o l u t i o n , and accuracy w i t h which image data can be gathered. In the C e l l Analyzer, a l i n e a r charge-coupled (CCD) sensor i s used f o r image d e t e c t i o n . A s o l i d s t a t e imaging device was s e l e c t e d over a v i d i c o n tube due to i t s s u p e r i o r s p a t i a l and photometric r e s o l u t i o n ( P a l c i c and J a g g i , 1986). Images are 29 e s s e n t i a l l y f r e e of geometric d i s t o r t i o n introduced by the d e t e c t o r , and high p o s i t i o n a l accuracy can be achieved. In a d d i t i o n , CCD detectors have a l i n e a r response over t h e i r e n t i r e dynamic range, so the absolute i n t e n s i t y of an image can be obtained d i r e c t l y from the de t e c t o r output s i g n a l (Hiraoka e t a l , 1987). Furthermore, use of a l i n e a r , r a t h e r than a 2-dimensional, s o l i d s t a t e array allowed mechanical scanning i n one d i r e c t i o n (the x - d i r e c t i o n ) to be combined w i t h e l e c t r o n i c scanning i n the per p e n d i c u l a r d i r e c t i o n (the y - d i r e c t i o n ) (Jaggi and P a l c i c , 1985). With proper adjustment of the mechanical scanning speed, s p a t i a l r e s o l u t i o n of the image co u l d be ad j u s t a b l e i n the scanning d i r e c t i o n , and each image l i n e c o u l d be c o l l e c t e d and processed as the sample was moved. By c o n t r a s t , a 2-dimensional array or v i d i c o n tube would r e q u i r e frame-by-frame scanning. This would r e s u l t i n a constant s e r i e s of a c c e l e r a t i o n s and d e c e l e r a t i o n s of the sample, s i n c e image data c o u l d not be c o l l e c t e d as the sample was moved. In a d d i t i o n , the data i n each frame would have to be processed before c a p t u r i n g the next frame, making r e a l time p r o c e s s i n g much more d i f f i c u l t to implement. Real time image pr o c e s s i n g was e s s e n t i a l because of the i m p r a c t i c a l l y l a r g e data storage c a p a b i l i t i e s t h a t would be r e q u i r e d f o r even a s i n g l e t i s s u e c u l t u r e f l a s k ( i . e . a 10 cm 2 area scanned at 1 u-m2 r e s o l u t i o n would r e q u i r e 1 gigabyte of storage space) . A 2-dimensional sensor would a l s o r e q u i r e the c o n s i d e r a t i o n of more image boundaries, and r e s o l u t i o n i n the x - d i r e c t i o n r e l a t i v e to r e s o l u t i o n i n the y - d i r e c t i o n would be f i x e d . Because of t h i s l a t t e r r e s t r i c t i o n , means of i n c r e a s i n g scanning speed would be l i m i t e d . The l i n e a r s o l i d s t a t e image sensor i n the C e l l Analyzer i s a 1728 p i x e l CCD array (CCD122, F a i r c h i l d ) mounted i n an image-scanning camera (612, Datacopy Co r p o r a t i o n ) . The p i x e l s i z e i s 13 nm X 13 //m. Arranged i n a l i n e a r a r r a y , 1728 such p i x e l s y i e l d a t o t a l sensor l e n g t h of 22.5 mm. 30 The area that could be scanned was t h e r e f o r e about twice as wide as the p i c t u r e area f o r a t y p i c a l v i d i c o n camera tube, and was, furthermore, obtained at 2-3 times higher s p a t i a l r e s o l u t i o n . The photometric r e s o l u t i o n of the sensor i s 256 i n t e n s i t y l e v e l s ("grey l e v e l s " ) , and the i n t e g r a t i o n time r e q u i r e d f o r l i g h t c o l l e c t i o n i s 3.5 msec. Adding a 3.5 msec image read-out time to the i n t e g r a t i o n time, the t o t a l scanning time f o r one image l i n e becomes 7 msec. 3.3 Image Generation 3.3.1 Image Formation i n the L i g h t Microscope The compound v i s i b l e l i g h t microscope i s a powerful t o o l f o r the viewing and imaging of b i o l o g i c a l specimens at the c e l l u l a r l e v e l . High q u a l i t y o p t i c a l components are r e a d i l y a v a i l a b l e , and can be e a s i l y mounted and a l i g n e d on the microscope base. In a d d i t i o n , specimens can be p r e c i s e l y p o s i t i o n e d and focused i n the o p t i c a l path of the instrument. The b a s i c components of the o p t i c a l path i n a compound v i s i b l e l i g h t microscope are: the i l l u m i n a t o r , the condenser l e n s , the o b j e c t i v e l e n s , and the o c u l a r or p r o j e c t i o n l e n s . The specimen to be imaged i s placed i n between the condenser and o b j e c t i v e lenses. The arrangement of these components i n an i n v e r t e d compound microscope i s i l l u s t r a t e d i n Figure 5a. Inverted microscopes a l l o w specimens to be viewed from below, a n e c e s s i t y i n t i s s u e c u l t u r e , where c e l l s must be viewed through comparatively bulky c o n t a i n e r s . As shown i n the f i g u r e , v a r i o u s diaphragms, prisms, and m i r r o r s are used i n a d d i t i o n to the b a s i c components i n order to f u r t h e r d e fine the l i g h t ray paths. I t should a l s o be noted t h a t , w h i l e the condenser, o b j e c t i v e , and o c u l a r are represented as s i n g l e lenses i n the f i g u r e , they are a c t u a l l y c onstructed as complex lens systems to produce 31 i l l u m i n a t o r o p t i c a l a x i s Figure 5. Inverted l i g h t microscopy. (a) o p t i c a l system of an inverted microscope, showing the positions of the c o l l e c t o r , condenser, and objective, and t h e i r associated diaphragms ( n c o U> Dcond> n o b j ) • (b) bright f i e l d microscopy under Koehler i l l u m i n a t i o n , showing the two sets of conjugate image planes (A1 , A 2 , A 3, and Flt F 2 ) . A 2 i s the front f o c a l plane of the condenser, and A 3 i s the back f o c a l plane of the objective. The rays shown are i l l u m i n a t i o n paths. 32 the d e s i r e d p r o p e r t i e s f o r m i c r o s c o p i c a l imaging ( i n c l u d i n g c o r r e c t i o n of var i o u s a b e r r a t i o n s common to s i n g l e l e n s e s ) . A l s o evident i n Figure 5 i s the f a c t that the i l l u m i n a t o r c o n s i s t s of m u l t i p l e components. Along w i t h the condenser, these components make up the i l l u m i n a t i n g system. Proper adjustment of the i l l u m i n a t i n g system i s very important f o r the generation of hi g h q u a l i t y images. The requirements f o r proper i l l u m i n a t i o n i n most v i s i b l e l i g h t microscopy a p p l i c a t i o n s are: the specimen should be i l l u m i n a t e d only over the re g i o n of observation; i l l u m i n a t i o n should be uniform w i t h a d j u s t a b l e i n t e n s i t y ; and the cone of i l l u m i n a t i o n should be v a r i a b l e i n order to f i l l a d e s i r e d p o r t i o n of the o b j e c t i v e ' s aperture (thus minimizing g l a r e ) (Rochow and Rochow, 1978; P l u t a , 1987). To meet these c r i t e r i a , Koehler i l l u m i n a t i o n i s employed i n most l i g h t microscopy a p p l i c a t i o n s (see Figure 5b). In Koehler i l l u m i n a t i o n , the image of the l i g h t source ( i . e . the lamp f i l a m e n t ) i s focused i n the plane of the condenser diaphragm ( n c o n d i n Figure 5), and the condenser lens i s adjusted to focus the image of the i l l u m i n a t o r ' s diaphragm ( n c o l l) i n the plane of the specimen. The lamp's f i l a m e n t i s out of focus i n the o b j e c t plane, r e s u l t i n g i n even i l l u m i n a t i o n of the f i e l d . Another property of Koehler i l l u m i n a t i o n i s that i t produces two sets of conjugate planes (planes which are i n focus r e l a t i v e to each other) that are r e c i p r o c a l l y r e l a t e d (see, f o r example, Inoue, 1986). L i g h t rays that are focused to a p o i n t i n one set of conjugate planes are e s s e n t i a l l y p a r a l l e l i n the other, and v i c e versa. The two sets of conjugate planes are l a b e l l e d Flt F 2 (the field planes) and k1, A 2, A 3 (the aperture planes) i n Figure 5b. The formation of m i c r o s c o p i c a l and other o p t i c a l images can be desc r i b e d i n terms of the Abbe theory of image formation. L i g h t i n c i d e n t on the ob j e c t plane i s d i f f r a c t e d by the specimen and focused by a lens ( i . e . the o b j e c t i v e ) i n t o a d i f f r a c t i o n spectrum. The d i f f r a c t i o n spectrum occurs i n the image plane of the i l l u m i n a t i n g l i g h t ( i . e . i n the plane conjugate to the l i g h t source, i f there are no lenses modifying the path of the l i g h t rays emanating from the source) (Hecht, 1987; Thompson, 1978; H u t z l e r , 1977). This plane i s commonly r e f e r r e d to as the F o u r i e r transform plane, s i n c e , i n o p t i c a l systems that are i l l u m i n a t e d c oherently, the d i f f r a c t i o n spectrum can be described by the F o u r i e r transform: F(u,v) f ( x , y ) exp[ -i27r(xu+yv) ] dxdy (3.1) where f ( x , y ) i s the amplitude transmittance of the specimen, and F(u,v) describes the d i s t r i b u t i o n of s p a t i a l frequencies i n f ( x , y ) (see Figure 6a) (Goodman, 1968). Because of i t s transforming p r o p e r t i e s , the imaging lens i s o f t e n r e f e r r e d to as a transform l e n s . I n t e r f e r e n c e of the wavefronts propagating beyond the F o u r i e r transform plane r e s u l t s i n the formation of an i n v e r t e d image of the specimen i n the image plane conjugate to the obje c t plane. The formation of t h i s image can a l s o be described as a F o u r i e r transform process. In t h i s case, the image of the specimen can be d e s c r i b e d as a F o u r i e r t r a n s f o r m a t i o n of the s p a t i a l frequency d i s t r i b u t i o n , F ( u , v ) , seen i n the F o u r i e r transform plane. Because of the f i n i t e nature of the components i n an o p t i c a l system, the s p a t i a l frequency spectrum produced i n the F o u r i e r transform plane i s ba n d - l i m i t e d . In other words, hig h frequency components are "cut o f f " by the f i n i t e aperture of the imaging l e n s . This l i m i t s the r e s o l u t i o n of the 34 ( 3, I object plane x 7 ' transform i v lens Fourier Figure 6. D i f f r a c t i o n and image r e s o l u t i o n i n an o p t i c a l system, (a) formation of the s p a t i a l frequency spectrum, F(u,v), of a l i g h t -absorbing object described by the amplitude transmittance function, f ( x , y ) . Under p a r a l l e l i l l u m i n a t i o n , the frequency spectrum occurs i n the back f o c a l plane of the lens (the Fourier plane). (b) image formation of a point source by a d i f f r a c t i o n - l i m i t e d o p t i c a l system. D i f f r a c t i o n from the f i n i t e lens aperture produces a point-spread (Airy) i n t e n s i t y pattern i n the image plane. d' i s the minimum distance that can separate two resolvable A i r y patterns ( a r i s i n g from separate point sources) according to the Rayleigh c r i t e r i o n . 35 o p t i c a l system, s i n c e i t i s the h i g h s p a t i a l frequencies t h a t produce the sharp d e t a i l i n an image. The l i m i t of r e s o l u t i o n of an o p t i c a l system depends on the c u t - o f f value f o r the h i g h s p a t i a l frequencies i n the F o u r i e r transform plane. In terms of the p h y s i c a l p r o p e r t i e s of the o p t i c a l components of a system, the minimum d i s t a n c e , d, between two p o i n t s i n an o b j e c t t h a t w i l l appear separated i n i t s image i s : d = A / n s i n a (3.2) where n i s the r e f r a c t i v e index of the medium through which the l i g h t rays are t r a v e l l i n g ( u s u a l l y a i r ) , A i s the wavelength of l i g h t i n a i r , and a i s the angle between the o p t i c a l a x i s and the most d i v e r g i n g rays t h a t enter the o b j e c t i v e and e v e n t u a l l y reach the image plane ( r e f e r to Figure 6b). I f the angle a i s determined by the aperture of the o b j e c t i v e lens i t s e l f (and not by some other element i n the o p t i c a l path) , the q u a n t i t y n s i n a i s r e f e r r e d to as the numerical aperture (NA) of the l e n s . In a microscope i l l u m i n a t e d according to the Koehler p r i n c i p l e , the r e s o l v i n g power can be expressed i n terms of the numerical apertures of the o b j e c t i v e and condenser lenses ( P l u t a , 1988): d = A / (NA + NA .) . (3.3) ' v obj cond' I t should be noted that Equation 3.2 i s based on the a b i l i t y to r e s o l v e elements i n a p e r i o d i c s t r u c t u r e i l l u m i n a t e d normally by coherent l i g h t . In the microscope, i l l u m i n a t i o n i s g e n e r a l l y incoherent or p a r t i a l l y coherent. Under these c o n d i t i o n s , r e s o l v i n g power i s commonly expressed i n terms of the Rayleigh c r i t e r i o n , d = 1.22 A / In s i n a. (3.4) This c r i t e r i o n i s based on the image formed by a s i n g l e luminous p o i n t through a lens of f i n i t e aperture. Because of d i f f r a c t i o n l i m i t a t i o n s , p o i n t sources are imaged as the well-known A i r y p a t t e r n s (see Figure 6b). Incoherently i l l u m i n a t e d objects can be considered as an a r r a y of non-i n t e r f e r i n g luminous p o i n t sources, and the R a y l e i g h c r i t e r i o n s t a t e s that two such p o i n t s can be r e s o l v e d i f the c e n t r a l maxima of t h e i r A i r y p a t t e r n s do not overlap ( i . e . i f the c e n t r a l maximum of one A i r y p a t t e r n f a l l s at or beyond the l o c a t i o n of the f i r s t minimum of i t s neighbour). As w i t h the g r a t i n g r e s o l u t i o n of Equation 3.2, the numerical aperture of the condenser lens under Koehler i l l u m i n a t i o n a f f e c t s the R a y l e i g h r e s o l u t i o n , y i e l d i n g , d = 1.22 A / (NA + NA . ) . (3.5) ' v obj cond 3.3.2 Imaging of Phase Objects Using S p a t i a l F i l t e r i n g Techniques Since l i v e c e l l s are e s s e n t i a l l y transparent, they produce images w i t h l i t t l e or no c o n t r a s t under b r i g h t f i e l d i l l u m i n a t i o n . However, they g e n e r a l l y have a r e f r a c t i v e index d i f f e r e n t from that of the medium i n which they grow. Such objects are r e f e r r e d to as phase o b j e c t s , because they a f f e c t the phase, but not the amplitude, of the l i g h t that passes through or i s d i f f r a c t e d by them. Because image detectors ( i n c l u d i n g the human eye) cannot detect phase v a r i a t i o n s , no v i s i b l e image i s produced. S p a t i a l f i l t e r i n g i n the F o u r i e r transform plane, can, however, be used to s e l e c t i v e l y modify l i g h t rays that have been d i f f r a c t e d or r e f r a c t e d by the o b j e c t . Background l i g h t that has not i n t e r a c t e d w i t h the specimen may a l s o be modified, attenuated, or blocked. Such s p a t i a l f i l t e r i n g methods 37 are c o l l e c t i v e l y r e f e r r e d to as S c h l i e r e n techniques. These techniques a l s o have widespread a p p l i c a t i o n s i n the d e t e c t i o n of a b e r r a t i o n s i n o p t i c a l elements (see, f o r example, Ojeda-Castaneda, 1978) and f o r determining r e f r a c t i v e i n d i c e s of minerals, chemicals, and other m a t e r i a l s (McCrone, 1975; Dodd and McCrone, 1975). One of the most well-known s p a t i a l f i l t e r i n g methods a p p l i e d to microscopy i s the phase contrast method, developed by Zernike i n 1934 (Zernike, 1934; Zernike, 1958). This method creates c o n t r a s t from a phase objec t by causing c o n s t r u c t i v e or d e s t r u c t i v e i n t e r f e r e n c e between unmodified background l i g h t rays and l i g h t rays d i f f r a c t e d by the obj e c t . As shown e a r l i e r i n Figure 6a, l i g h t d i f f r a c t e d by a specimen forms a s p a t i a l frequency d i s t r i b u t i o n i n the F o u r i e r transform plane of the imaging l e n s , w h i l e the u n d i f f r a c t e d background i l l u m i n a t i o n i s focused i n t o a b r i g h t c e n t r a l spot ( c a l l e d the zero t h order because i t has zero s p a t i a l frequency). Because the d i f f r a c t e d l i g h t c o n t r i b u t e s predominantly to the higher order s p a t i a l frequencies ( r a t h e r than to the ze r o t h o r d e r ) , the p o s i t i o n i n g of an a p p r o p r i a t e l y constructed p h a s e - a l t e r i n g f i l t e r i n the F o u r i e r transform plane can s e l e c t i v e l y s h i f t the phase of e i t h e r the d i f f r a c t e d or the background l i g h t . Figure 7 i l l u s t r a t e s the o p t i c a l arrangement used to generate a phase c o n t r a s t image. The f i l t e r p laced i n the F o u r i e r transform plane c o n s i s t s of a f l a t g l a s s p l a t e made up of a c e n t r a l , p a r t i a l l y absorbing mask surrounded by a s l i g h t l y t h i c k e r (but transparent) annular r e g i o n . The c e n t r a l mask acts to reduce the i n t e n s i t y of the unmodulated background l i g h t to a l e v e l comparable to that of the d i f f r a c t e d l i g h t , w h i l e the outer annulus allows the d i f f r a c t e d l i g h t to pass through without a p p r e c i a b l e a t t e n u a t i o n . Because the outer annulus i s s l i g h t l y t h i c k e r than the c e n t r a l mask, the d i f f r a c t e d l i g h t s u f f e r s a greater phase Figure 7. Generation of a phase contrast image under p a r a l l e l incident i l l u m i n a t i o n . In a microscope, the transform lens would be the objective. 39 r e t a r d a t i o n t h a n t h e b a c k g r o u n d l i g h t . The amount o f phase r e t a r d a t i o n n e c e s s a r y t o p r o d u c e d e s t r u c t i v e i n t e r f e r e n c e between t h e d i f f r a c t e d and b a c k g r o u n d r a y s i n t h e image p l a n e was f o u n d by Z e r n i k e t o be one q u a r t e r o f a w a v e l e n g t h ( i n h i s e x p e r i m e n t s , Z e r n i k e f o u n d t h a t l i g h t waves d i f f r a c t e d by t h i n , t r a n s p a r e n t specimens were r e t a r d e d by a q u a r t e r w a v e l e n g t h , so an a d d i t i o n a l q u a r t e r - w a v e r e t a r d a t i o n i n t h e phase p l a t e w o u l d l e a d t o d e s t r u c t i v e i n t e r f e r e n c e ) . A d a r k image a g a i n s t a b r i g h t b a c k g r o u n d i s p r o d u c e d . T h i s i s r e f e r r e d t o as p o s i t i v e phase c o n t r a s t . N e g a t i v e phase c o n t r a s t ( b r i g h t image on d a r k e r b a c k g r o u n d ) i s p r o d u c e d i f c o n s t r u c t i v e i n t e r f e r e n c e i s i n d u c e d by making t h e c e n t r a l mask o f the phase p l a t e t h i c k e r t h a n t h e o u t e r a n n u l u s . I n p r a c t i c e , phase c o n t r a s t m i c r o s c o p y i s p e r f o r m e d under K o e h l e r i l l u m i n a t i o n , s i n c e t h e arrangement i l l u s t r a t e d i n F i g u r e 7 i s e q u i v a l e n t t o a c o n d e n s e r w i t h z e r o n u m e r i c a l a p e r t u r e . Image r e s o l u t i o n i s t h e r e f o r e n o t o p t i m a l (see E q u a t i o n 3.5). The phase c o n t r a s t s e t - u p u s e d w i t h K o e h l e r i l l u m i n a t i o n i s i l l u s t r a t e d i n F i g u r e 8a. R a t h e r t h a n p l a c i n g a s i n g l e s p a t i a l f i l t e r i n the F o u r i e r t r a n s f o r m p l a n e , a c o m b i n a t i o n o f an a n n u l a r d i a p h r a g m i n t h e f r o n t f o c a l p l a n e o f t h e c o n d e n s e r and an a n n u l a r phase p l a t e i n t h e b a c k f o c a l p l a n e o f t h e o b j e c t i v e i s u s e d ( i t s h o u l d be n o t e d t h a t t h e phase p l a t e i s g e n e r a l l y l o c a t e d w i t h i n t h e o b j e c t i v e ; s p e c i a l phase c o n t r a s t o b j e c t i v e s t h e r e f o r e have t o be u s e d f o r phase c o n t r a s t m i c r o s c o p y ) . The above arrangement c a u s e s t h e s p ecimen t o be i l l u m i n a t e d w i t h a h o l l o w cone o f i n c i d e n t l i g h t . Because t h e b a c k f o c a l p l a n e o f t h e o b j e c t i v e i s c o n j u g a t e t o t h e f r o n t f o c a l p l a n e o f the c o n d e n s e r ( c f . F i g u r e 5 b ) , the u n d i f f r a c t e d r a y s f r o m t h e i n c i d e n t l i g h t p a s s t h r o u g h an a n n u l a r r e g i o n o f t h e o b j e c t i v e ' s b a c k f o c a l p l a n e . T h i s r e g i o n o f t h e phase p l a t e i s t h e r e f o r e a n a l o g o u s t o the c e n t r a l mask o f the c o n f i g u r a t i o n i n F i g u r e 7. D i f f r a c t e d r a y s s t r i k e p r i m a r i l y t h e o t h e r 40 Figure 8. Phase contrast and modulation contrast microscopy. (a) generation of a phase contrast image under Koehler i l l u m i n a t i o n . (b) modulation contrast microscopy. The p o l a r i z e r - s l i t combination (shown i n both side and plan views) i s placed i n plane A 2, while the modulator i s placed i n plane A 3. The image of the s l i t i s superimposed onto the plan view of the modulator. 41 p a r t s of the phase p l a t e , a l l o w i n g a phase s h i f t to be induced between the d i f f r a c t e d and i l l u m i n a t i n g rays. P o s i t i v e or negative phase c o n t r a s t can be generated, depending on which regions of the phase p l a t e are made t h i c k e r . Other s p a t i a l f i l t e r i n g techniques a p p l i e d to microscopy i n c l u d e the d a r k - f i e l d methods. In the c e n t r a l d a r k - f i e l d method, the c e n t r a l mask of Figure 7a (or the p a r t i a l l y absorbing annulus of Figure 7b) i s made completely opaque, so that a l l background i l l u m i n a t i o n i s f i l t e r e d out. A b r i g h t image (due to the d i f f r a c t e d rays) on a completely dark background r e s u l t s . S i m i l a r r e s u l t s are obtained by i l l u m i n a t i n g the specimen o b l i q u e l y so th a t no background l i g h t enters the o b j e c t i v e . L i g h t i n t e n s i t i e s i n the F o u r i e r transform plane are a l s o sometimes s e l e c t i v e l y modulated, as opposed to b l o c k i n g out e n t i r e s p a t i a l frequency orders. In modulation contrast microscopy (Hoffman and Gross, 1975), t h i s modulation i s achieved u s i n g v a r i a b l e c r o s s i n g of p o l a r i z e r s . An o f f -c e n tre, v a r i a b l e - w i d t h s l i t diaphragm i s placed i n the f r o n t f o c a l plane of the condenser, r e s u l t i n g i n oblique i l l u m i n a t i o n of the specimen. A v a r i a b l e p a r t of the s l i t i s covered by a p o l a r i z e r . A l s o i n f r o n t of the condenser i s a second, r o t a t a b l e p o l a r i z e r (see Figure 8b) that allows modulation of l i g h t i n t e n s i t i e s through the p o r t i o n of the s l i t covered by the other p o l a r i z e r . In the back f o c a l plane of the o b j e c t i v e , the s l i t image i s focused onto a modulator w i t h dark, grey, and b r i g h t segments. This f o c u s i n g i s done i n such a way that the p o l a r i z e d p a r t of the s l i t image f a l l s onto the b r i g h t r e gion of the modulator, whereas the b r i g h t r e g i o n of the s l i t image f a l l s onto the grey r e g i o n of the modulator. Using such a system, the i n t e n s i t y of d i f f e r e n t p o r t i o n s of the frequency spectrum can be modulated. In p a r t i c u l a r , opposite phase gradients i n the obj e c t can be converted to opposite i n t e n s i t i e s i n the image, r e s u l t i n g i n 42 a form of o p t i c a l shading that produces images w i t h a 3-dimensional appearance. 3.3.3 Evaluation of S p a t i a l F i l t e r i n g Methods f o r C e l l Detection Purposes B r i g h t f i e l d , phase c o n t r a s t , and dark f i e l d microscopy modes were t e s t e d f o r t h e i r a p p l i c a b i l i t y to l i v e c e l l d e t e c t i o n procedures. Photographs of V79 c e l l images obtained under these microscopy modes are shown i n Figure 9. A l l images are of the same group of c e l l s , and were taken at low m a g n i f i c a t i o n (4.0 and 6.3X o b j e c t i v e m a g n i f i c a t i o n ) , which must be used i f r a p i d scanning i s to be achieved. I t i s evident from the photographs t h a t the V79 c e l l s generate very l i t t l e c o n t r a s t under b r i g h t f i e l d i l l u m i n a t i o n , as expected (Figure 9a). The phase c o n t r a s t image i s shown i n Figure 9b, and i s c h a r a c t e r i z e d by b r i g h t halos surrounding each c e l l . This i s a common feature of phase c o n t r a s t images, and presents some disadvantages because the true dimensions of the o b j e c t s are obscured. C e n t r a l d a r k - f i e l d images are shown i n Figures 9c,d f o r two d i f f e r e n t m a g n i f i c a t i o n s . The image i s very dim at 6.3X m a g n i f i c a t i o n , but the c e l l s are d i s t i n c t i f a 4.OX o b j e c t i v e of the same numerical aperture i s used. A l a r g e number of extraneous objects i s a l s o evident i n the dark f i e l d images. This was not n o t i c e a b l e under the other microscopy modes, and may have been caused by debris or by imperfections i n the p l a s t i c of the t i s s u e c u l t u r e v e s s e l . The presence of such a l a r g e amount of d e t e c t a b l e debris would be very disavantageous f o r object d e t e c t i o n and c e l l r e c o g n i t i o n purposes. L i n e a r CCD sensor scans of c e l l images i n the b r i g h t f i e l d and phase c o n t r a s t microscopy modes are shown i n Figure 10. A s i n g l e l i n e scan ( c r o s s i n g two of the three c e l l s p i c t u r e d i n Figure 9), as w e l l as a 2-dimensional i n t e n s i t y p l o t generated from successive l i n e scans taken at Figure 9. V79 c e l l images produced by d i f f e r e n t microscopy modes. Top l e f t : b r i g h t f i e l d , 6.3/0.16 (6.3X magnification, 0.16 numerical aperture) objective. Top r i g h t : phase contrast, 6.3/0.16 objective. Bottom l e f t : d a r k - f i e l d , 6.3/0.16 objective. Bottom r i g h t : d a r k - f i e l d , 4.0/0.16 objective. 44 200 13 150 > SolOO .1? 50 (a) bright field 0 L ; (b) phase contrast Figure 10. Image detection by a l i n e a r CCD sensor under bri g h t f i e l d and phase contrast microscopy modes. Single l i n e scans as well as 2-dimensional i n t e n s i t y plots are shown. The c e l l s imaged are the same as those i n Figure 9. A 6.3/0.16 objective i n combination with a 2. 5X p r o j e c t i o n lens was used to produce both sets of images. 45 1 m^ i n t e r v a l s , i s shown f o r each microscopy mode. A sm a l l amount of c o n t r a s t due to absor p t i o n could be achieved i n the b r i g h t f i e l d mode w i t h c a r e f u l adjustment of the focus (Figure 10a) ( i t should be noted, however, th a t t h i s c o n t r a s t i s achieved when c e l l s are s l i g h t l y out of focus, r a t h e r than p e r f e c t l y i n focus, where c o n t r a s t almost completely disappears). At i t s best, the c o n t r a s t that could be generated i n the b r i g h t f i e l d mode spanned only about 20 grey l e v e l s out of the 256 grey l e v e l photometric range of the l i n e a r sensor. The photometric i n f o r m a t i o n t h a t c o u l d be obtained from the images was th e r e f o r e minimal, and was p a r t i c u l a r l y u n s u i t a b l e f o r d e t e c t i o n purposes, where l a r g e s i g n a l s are advantageous. Sensor scans of the phase c o n t r a s t images are shown i n Figure 10b. These, too, spanned only a small (20-30) grey l e v e l range, although t h i s might have been improved w i t h an o v e r a l l increase i n the image i n t e n s i t y . However, standard microscope l i g h t sources are not a v a i l a b l e w i t h power r a t i n g s h i gher than 100 watts, which would be r e q u i r e d i f s i g n i f i c a n t i ncreases i n i n t e n s i t y were to be achieved i n the phase c o n t r a s t mode (the image of Figure 10b was obtained w i t h 85 watt i l l u m i n a t i o n ; t h i s can be compared w i t h 25 watts f o r the b r i g h t f i e l d mode). Despite the low image i n t e n s i t y , however, the phase c o n t r a s t mode d i d appear to produce a c h a r a c t e r i s t i c c e l l s i g n a t u r e . Dark f i e l d images were not detectable w i t h the l i n e a r sensor, even w i t h 90 watt i l l u m i n a t i o n . They may have been de t e c t a b l e i f the i n t e g r a t i o n time of the sensor had been increased, but t h i s would have r e s u l t e d i n s i g n i f i c a n t l y increased scanning times and CCD dark c u r r e n t s . This microscopy mode was th e r e f o r e r u l e d out f o r c e l l d e t e c t i o n purposes, having a l s o the disadvantage of enhancing s i g n a l s from l a r g e numbers of n o n - c e l l o b j e c t s . 46 3.3.4 Generation of C e l l S i g n a l s v i a the "Lens" E f f e c t Because of the low i n t e n s i t y (and low photometric range) of c e l l images obtained by s p a t i a l f i l t e r i n g methods, a l t e r n a t i v e means of generating s i g n a l s from l i v e c e l l s were sought. C o n s i d e r a t i o n of the s p h e r i c a l shape of f r e s h l y p l a t e d c e l l s l e d to the hypothesis that they might behave o p t i c a l l y l i k e t i n y lenses ( P a l c i c and J a g g i , 1986). I f exposed to p a r a l l e l , r a t h e r than Koehler, i l l u m i n a t i o n , they might t h e r e f o r e focus the incoming l i g h t to form a b r i g h t spot at a distance e q u i v a l e n t to one " f o c a l length" away. I n v e s t i g a t i o n of t h i s p o s s i b i l i t y showed that t h i s phenomenon d i d indeed e x i s t , and i t was termed the "lens e f f e c t " . Approximately p a r a l l e l i l l u m i n a t i o n of the o b j e c t plane i n the v i c i n i t y of the CCD detector was achieved by removing the condenser from the microscope i l l u m i n a t i o n system. The c o n f i g u r a t i o n used i s shown i n Figure 11, and w i l l be r e f e r r e d to as the modified b r i g h t f i e l d mode since no s p a t i a l f i l t e r i n g was i n v o l v e d . C e l l images produced w i t h t h i s c o n f i g u r a t i o n are shown i n Figure 12. The c e l l s i n these photographs are the same as those seen i n Figure 9. The modified b r i g h t f i e l d mode produces c e l l images c h a r a c t e r i z e d by a b r i g h t c e n t r a l r e g i o n o u t l i n e d by a dark r i n g . These same features can a l s o be c l e a r l y seen i n l i n e scans detected by the l i n e a r CCD sensor (see Figure 13) . A s i n g l e l i n e scan across a c e l l produces a d i s t i n c t c e l l s ignature c h a r a c t e r i z e d by a b r i g h t c e n t r a l maximum, or peak, bordered by a dark r i n g . Outside the dark r i n g i s a low i n t e n s i t y secondary maximum, which w i l l be r e f e r r e d to as the " b r i g h t rim". In a d d i t i o n to the production of a c h a r a c t e r i s t i c c e l l s i g n a t u r e , the m o d i f i e d b r i g h t f i e l d mode y i e l d e d s i g n a l s w i t h s u f f i c i e n t l y h i g h i n t e n s i t y and c o n t r a s t to e x p l o i t almost the e n t i r e photometric range of the l i n e a r 47 light source collector diffuser image plane of light source ^ approximately parallel light at object plane cell image plane Figure 11. Image formation i n the modified b r i g h t f i e l d microscopy mode. The l i g h t source i s adjusted such that the c o l l e c t o r image of the lamp fi l a m e n t i s f a r away from the specimen. A l m o s t - p a r a l l e l l i g h t then i l l u m i n a t e s the c e n t r a l p o r t i o n of the object plane. 4 8 1 1 o o o o Figure 12. V79 c e l l images produced by the mo d i f i e d b r i g h t f i e l d microscopy mode. L e f t : 6.3/0.16 o b j e c t i v e . R i g h t : 4.0/0.20 o b j e c t i v e (note: the dark v e r t i c a l l i n e at the top of the p i c t u r e i s a p o r t i o n of the microscope c r o s s - h a i r ) . The c e l l s are the same as those shown i n Figure 9. Figure 13. Image detection by a l i n e a r CCD sensor under the modified b r i g h t f i e l d microscopy mode. Both single l i n e scans and 2-dimensional i n t e n s i t y p l o t s are shown. The c e l l s imaged are the same as those i n Figures 9, 10, and 12. A 2. 5X p r o j e c t i o n lens was used to produce both sets of images. 50 sensor. This was achieved with only 35-45 watt incident i l l u m i n a t i o n , despite the f a c t that the incident l i g h t was not focused into an intense cone at the object plane. Furthermore, any image r e s o l u t i o n that was l o s t by changing from Koehler i l l u m i n a t i o n to p a r a l l e l i l l u m i n a t i o n was of l i t t l e consequence since only low magnification (and thus low s p a t i a l resolution) images were being generated. Further investigations revealed that only a small proportion of the debris i n a t y p i c a l tissue culture preparation would focus the incoming l i g h t s u f f i c i e n t l y to produce an above-background s i g n a l . Thus, a large proportion of debris would be ignored by a detection procedure that sought only signals with i n t e n s i t i e s greater than a c e r t a i n l e v e l above background. The modified b r i g h t f i e l d mode was therefore selected over the s p a t i a l f i l t e r i n g techniques for the generation of images for c e l l detection and recognition purposes. 3.4 Overview of System Hardware A photograph of the C e l l Analyzer i s shown i n Figure 14, and a block diagram of the system hardware i s shown i n Figure 15. A Nikon Diaphot inverted microscope set up i n the modified b r i g h t f i e l d microscopy mode of Figure 11 was used to generate a l l the images for the purposes of t h i s t h e s i s . To ensure stable i l l u m i n a t i o n of the specimen i n t h i s mode, a DC power driven quartz halogen l i g h t source was used. The Datacopy 612 camera containing the F a i r c h i l d CCD122 array was mounted i n the camera port of the microscope (see Figure 14) . E f f e c t i v e s p a t i a l r e s o l u t i o n of the sensor could be adjusted by varying the microscope magnification. Image processing was c a r r i e d out i n r e a l time by a TMS 32010 d i g i t a l s i g n a l 51 Figure 14. The C e l l Analyzer system. Top: photograph of system showing the Nikon Diaphot inverted microscope, the Datacopy 612 camera inserted into the microscope camera port, the j o y s t i c k f o r manual control of microscope stage movements, and the IBM AT host computer. Half of the custom-fitted incubator i s also i n place. Bottom: close-up photograph of a 25 cm2 Nunclon tissue culture f l a s k positioned on the microscope stage. 52 INVERTED MICROSCOPE 7T Z-DRIVE TISSUE CULTURE VESSEL PRECISION X,Y MICROSCOPE STAGE Z-DRIVE CONTROLLER JOYSTICK X,Y STAGE CONTROLLER •7 ) DIGITAL SCANNER (LINEAR SOLID STATE IMAGE SENSOR) DIGITAL SIGNAL PROCESSOR AND MULTIPORT MEMORY IBM A T HOST COMPUTER Figure 1 5 . Block diagram of the C e l l Analyzer. 53 processor (Texas Instruments), which communicated d i r e c t l y w i t h the l i n e a r sensor through a m u l t i p o r t memory. Three-dimensional scanning of the t i s s u e c u l t u r e f l a s k was c a r r i e d out by a p r e c i s i o n x,y microscope stage (Maerzhaeuser) and focus d r i v e ( r e f e r r e d to as the z - d r i v e ) . Tissue c u l t u r e specimens were f i r m l y and r e p r o d u c i b l y seated on the stage u s i n g a custom-made holder (see Figure 14b). The x,y stage could be moved i n 1 fxm steps at a maximum speed of 20,000 steps per second by two p r e c i s i o n stepping motors. Scanning of the f l a s k was achieved by moving the stage i n the x - d i r e c t i o n , which i s p e r p e n d i c u l a r to the o r i e n t a t i o n of the l i n e a r sensor (see Figure 16). Focus c o u l d be adjusted using the z - d r i v e , which had a stepping motor capable of moving the f o c u s i n g mechanism of the microscope i n 0.5 /zm increments. High r e s o l u t i o n scans could be made at h i g h m a g n i f i c a t i o n by mechanically moving the l i n e a r sensor i t s e l f . The sensor c o u l d be moved over a distance of 37 mm, although much smaller d i s t a n c e s were u s u a l l y used. The sensor was moved i n 13 /j.m steps, which matched the p i x e l s i z e of the l i n e a r array. The r e s o l u t i o n of the f i n e scan i n both the x- and y - d i r e c t i o n s was t h e r e f o r e the same, and depended on the image m a g n i f i c a t i o n . In c o n t r a s t , i f the microscope stage was used f o r stepping, the r e s o l u t i o n i n the x - d i r e c t i o n was determined by the a c t u a l step s i z e , whereas the r e s o l u t i o n i n the y - d i r e c t i o n depended on the m a g n i f i c a t i o n used. An IBM AT host computer ( I n t e l 80286/287) was the system c o n t r o l l e r and provided f o r i n t e r a c t i o n w i t h the experimenter through v a r i o u s software programs. During C e l l Analyzer operation, the host computer i n i t i a l i z e d the system and then managed the operations of the system components. The 54 Figure 16. Scanning procedure for c e l l detection. The f l a s k i s scanned by moving i t across the l i n e a r sensor using the computer-controlled microscope stage. Sensor scans are taken at s p e c i f i e d i n t e r v a l s as the f l a s k i s moved. 55 components were operated i n p a r a l l e l by t h e i r own dedicated microprocessors. The computer was also used for data storage and analysis. In order to allow time-lapse studies of mammalian c e l l behavior to be c a r r i e d out at appropriate environmental temperatures, the C e l l Analyzer was f i t t e d with a t i n t e d , removable Ple x i g l a s incubator that enclosed the microscope and stage (ha l f of t h i s incubator i s i n place i n the'photograph of Figure 14; the other h a l f had been removed to make the system components more v i s i b l e ) . A Nikon incubator-blower with thermoresistor feedback was used to c o n t r o l the a i r temperature inside the incubator. 3.5 System Software Several software programs have been generated by various users and developers of the C e l l Analyzer. These programs d i r e c t the performance of various tasks involved i n the detection, recognition, and quantitative c h a r a c t e r i z a t i o n of l i v e c e l l s . The programs relevant to t h i s thesis are described i n the following: CSC AN: This program (Jaggi et a l , 1986) was used to gather the i n i t i a l data set for an experiment. I t coordinated the search f o r c e l l s i n a s p e c i f i e d area of the tissue culture f l a s k , and recorded c e l l locations i n a data f i l e . The c e l l search involved a l i n e by l i n e scan, where the stage was moved a s p e c i f i e d i n t e r v a l between scans, as i l l u s t r a t e d i n Figure 16. The search procedure was c a r r i e d out at a t o t a l magnification of 10X (4.OX objective, 2. 5X p r o j e c t i o n lens), allowing a 2.25 mm wide band to be scanned with a si n g l e sweep of the sensor. A complete search of a 10 cm2 area i n a f l a s k involved 11 adjacent bands, each of which was 4 cm long. 56 Focus co u l d be adjusted a u t o m a t i c a l l y during the scanning procedure u s i n g procedures developed as a p a r t of t h i s t h e s i s . I f r e c o g n i t i o n parameters based on the s i g n a l features c a l c u l a t e d by the CSCAN program were a v a i l a b l e f o r the c e l l l i n e being s t u d i e d , the program would a l s o perform c e l l r e c o g n i t i o n as the f l a s k was scanned. An e x c l u s i o n procedure to remove from the data set those objects that were too c l o s e together was performed a f t e r the search, unless otherwise s p e c i f i e d by the experimenter. E x c l u s i o n of near neighbours was necessary to prevent confusion i n i d e n t i f y i n g the s e l e c t e d c e l l s when f u r t h e r measurements were made during subsequent r e v i s i t s . Objects remaining i n the data s et a f t e r e x c l u s i o n c o u l d be r e v i s i t e d f o r manual c l a s s i f i c a t i o n u s i n g CSCAN, but no f u r t h e r measurements were made on the c e l l s by t h i s program. In the r e v i s i t mode, each detected c e l l or object was moved to the center of the microscope f i e l d , and had to be c l a s s i f i e d by the experimenter before the next object was r e v i s i t e d ( u s u a l l y , a number between 0 and 9 was used to c l a s s i f y o b j e c t s ) . The experimenter had the op t i o n of r e v i s i t i n g a l l p o s t - e x c l u s i o n o b j e c t s , or only those that had been c l a s s i f i e d as c e l l s by the r e c o g n i t i o n a l g o r i t h m . Objects f o r which l o c a t i o n data was to be s t o r e d c o u l d be s e l e c t e d according to e i t h e r the manual or the machine c l a s s i f i c a t i o n s . CELREC: The purpose of t h i s program (Poon et a l , 1987) was to a i d i n the development of r e c o g n i t i o n algorithms to d i s t i n g u i s h c e l l s from d e b r i s detected i n the t i s s u e c u l t u r e f l a s k . The program allowed c e l l s i g n a l data ( i n c l u d i n g v a r i o u s c a l c u l a t e d s i g n a l features) to be s t o r e d and di s p l a y e d . C e l l s c o u l d be l o c a t e d e i t h e r manually by usin g a j o y s t i c k to move the microscope stage to s e l e c t e d l o c a t i o n s , or a u t o m a t i c a l l y by the same d e t e c t i o n procedures as i n CSCAN. Scanning parameters, such as step s i z e , stage speed, and s i z e of area scanned, could be modifi e d from w i t h i n the 57 program. I n d i v i d u a l c e l l s could be r e v i s i t e d not only f o r manual c l a s s i f i c a t i o n , but a l s o f o r more d e t a i l e d measurements, i n c l u d i n g determination of optimum focus, e v a l u a t i o n of feat u r e values at optimum focus and at v a r i o u s o f f s e t s from optimum focus, and measurement of feature values a t s e l e c t e d distances from the c e l l center. A s p e c i a l r e v i s i t i n g mode th a t simulated the scanning procedure, but paused wherever an objec t was detected, was a l s o a v a i l a b l e . Undetected o b j e c t s c o u l d be v i s u a l l y i d e n t i f i e d i n t h i s r e v i s i t i n g mode. SSCAN: This program ( P a l c i c and J a g g i , 1990) was used to r e v i s i t p r e v i o u s l y determined c e l l l o c a t i o n s a f t e r the c e l l s have been allowed to grow i n t o c o l o n i e s . SSCAN allowed manual c l a s s i f i c a t i o n of c o l o n i e s , and would a l s o a u t o m a t i c a l l y scan a re g i o n around each parent c e l l l o c a t i o n i n order to c a l c u l a t e a set of "colony f e a t u r e s " . Each of these scans c o n s i s t e d of 64 l i n e scans spaced 4 um apart by moving the microscope stage. Image data from the middle 256 p i x e l s of the l i n e a r sensor were analyzed (thus, a t 10X m a g n i f i c a t i o n , the e f f e c t i v e s i z e of the area scanned was 256 nm X 333 pm) . Coarse features were c a l c u l a t e d from each l i n e and summed f o r a l l the l i n e s to y i e l d o v e r a l l f e a t u r e scores f o r each colony. Feature scores could be st o r e d f o r f u r t h e r a n a l y s i s . RSCAN: This program (Thurston et a l , 1988) was developed p r i m a r i l y f o r r e v i s i t i n g s i n g l e c e l l s that were s e l e c t e d e i t h e r manually ( w i t h i n RSCAN) or a u t o m a t i c a l l y (using CSCAN). An automated r e v i s i t i n g mode cou l d r e v i s i t c e l l s repeatedly a t set time i n t e r v a l s , w i t h c e l l l o c a t i o n s being updated at each r e v i s i t through the automatic determination of the coordinates of the c e l l center. Various c e l l shape parameters, such as area, length, width, and b r i g h t n e s s , could a l s o be c a l c u l a t e d and s t o r e d during the 58 automatic r e v i s i t s . The images from which the parameters were c a l c u l a t e d were obtained by moving the stage i n 1 /xm steps i n the v i c i n i t y of the most r e c e n t l y recorded c e l l l o c a t i o n . For most purposes, a 96 X 96 p i x e l ( i . e . 96 stage steps u s i n g the middle 96 p i x e l s of the l i n e a r sensor) image was processed to o b t a i n these parameters. A l l the pro c e s s i n g was done i n r e a l time by the d i g i t a l s i g n a l processor as the microscope stage moved from one c e l l l o c a t i o n to the next. The 96 X 96 p i x e l raw images of s e l e c t e d c e l l s c o u l d be s t o r e d f o r f u r t h e r a n a l y s i s . Other r e v i s i t i n g options were a l s o a v a i l a b l e i n RSCAN. Objects could be manually c l a s s i f i e d , contour p l o t s of c e l l images co u l d be d i s p l a y e d , and colony f e a t u r e scores could be c a l c u l a t e d i n the same manner as i n SSCAN, but w i t h the c o l o n i e s being r e v i s i t e d repeatedly at s p e c i f i e d time i n t e r v a l s . 59 4. OPTIMIZATION OF CELL DETECTION 4.1 Criteria for Optimization In order to ensure that a minimum number of c e l l s was l e f t undetected when a f l a s k was scanned u s i n g the CSCAN or CELREC programs, f a c t o r s a f f e c t i n g c e l l s i g n a l q u a l i t y had to be i d e n t i f i e d and c o n t r o l l e d . For c e l l d e t e c t i o n purposes, the primary c e l l s i g n a l f e a t u r e t h a t had to be preserved was the s i g n a l peak. The l a r g e r the amplitude of the s i g n a l peak, the more e a s i l y a c e l l i s detected and d i s t i n g u i s h e d from other objects i n the f l a s k . Factors such as focus s e t t i n g , m a g n i f i c a t i o n , numerical aperture of the o b j e c t i v e l e n s , c e l l shape, and r e g i o n of the c e l l from which the s i g n a l i s obtained can s i g n i f i c a n t l y a f f e c t s i g n a l i n t e n s i t y . Acceptable v a r i a t i o n s due to these f a c t o r s had to be e s t a b l i s h e d , as d i d procedures to maintain the v a r i a t i o n s w i t h i n these l i m i t s . In a d d i t i o n to f a c t o r s a f f e c t i n g s i g n a l q u a l i t y , the scanning speed had to be considered i n the o p t i m i z a t i o n of the c e l l d e t e c t i o n procedure. Since s u r v i v a l experiments r e q u i r e l a r g e numbers of f l a s k s to be searched w i t h i n the span of a few hours, maximizing the scanning speed was of consid e r a b l e p r a c t i c a l importance. While the microscope stage can be d r i v e n a di s t a n c e of 1 pm i n as l i t t l e as 0.05 msec, 7 msec are r e q u i r e d to sense and process each image l i n e . Thus, the scanning speed was l i m i t e d by the d e t e c t i o n and processing time r a t h e r than by the speed l i m i t a t i o n s of the microscope stage. The only way of maximizing the scanning speed was to d r i v e the stage as large a d i s t a n c e , or "step", as p o s s i b l e between successive l i n e scans. 60 4.2 The C e l l D e t e c t i o n A l g o r i t h m The d e t e c t i o n a l g o r i t h m d e v e l o p e d f o r t h e l o c a t i o n o f o b j e c t s i n a t i s s u e c u l t u r e f l a s k ( J a g g i e t a l , 1986) made use o f t h e c h a r a c t e r i s t i c b r i g h t peak p r o d u c e d by c e l l s due t o t h e " l e n s e f f e c t " . S i g n a l peaks e x c e e d i n g a s p e c i f i e d above-background t h r e s h o l d i n t e n s i t y were s e a r c h e d f o r on each image l i n e . A t h r e s h o l d o f 15 g r e y l e v e l s above b a c k g r o u n d was u s u a l l y u s e d , s i n c e i t was l a r g e enough t o p r e v e n t b a c k g r o u n d n o i s e from b e i n g m i s t a k e n as a s i g n a l , b u t s m a l l enough so t h a t most c e l l s c o u l d be d e t e c t e d even under s u b - o p t i m a l c o n d i t i o n s , s u c h as p o o r l y a d j u s t e d f o c u s . Once a peak had been l o c a t e d on an image l i n e , f u r t h e r s i g n a l p r o c e s s i n g was c a r r i e d o u t t o d e t e r m i n e whether m u l t i p l e s i g n a l s had been d e t e c t e d f r o m t h e same o b j e c t . M u l t i p l e peaks c o u l d o c c u r a l o n g t h e same image l i n e , o r a t a p p r o x i m a t e l y the same p i x e l l o c a t i o n on a d j a c e n t l i n e s . C e l l d o u b l e t s ( i . e . c e l l s t h a t had d i v i d e d s i n c e b e i n g t r y p s i n i z e d and p l a t e d ) and some t y p e s o f d e b r i s c o u l d produce two o r more peaks on t h e same l i n e . These o b j e c t s , a l o n g w i t h s i n g l e c e l l s , c o u l d a l s o cause s i g n a l s t o be d e t e c t e d on a d j a c e n t image l i n e s . When m u l t i p l e s i g n a l s were d e t e c t e d w i t h i n a c o n f i n e d r e g i o n o f sp a c e , t h e c e l l d e t e c t i o n a l g o r i t h m r e c o r d e d them e i t h e r as s i n g l e o r as m u l t i p l e " h i t s " , d e p e n d i n g on c e r t a i n c o n d i t i o n s . I n t h e x - d i r e c t i o n ( d i r e c t i o n o f s c a n n i n g ) , s i g n a l s d e t e c t e d a t t h e same a p p r o x i m a t e p i x e l l o c a t i o n on l e s s t h a n 4 a d j a c e n t image l i n e s were r e c o r d e d as a s i n g l e " h i t " . O t h e r w i s e , 1 " h i t " was r e c o r d e d f o r e v e r y 4 a d j a c e n t l i n e s on w h i c h an o b j e c t was d e t e c t e d . I n t h e y - d i r e c t i o n , m u l t i p l e peaks a p p e a r i n g on t h e same image l i n e were i n t e r p r e t e d as b e l o n g i n g t o a s i n g l e o b j e c t i f t h e y f e l l w i t h i n 32 p i x e l s o f each o t h e r on t h e l i n e a r a r r a y . 61 4.3 O p t i m i z a t i o n of C e l l Shape While c e l l morphology v a r i e s f o r c e l l l i n e s o r i g i n a t i n g from d i f f e r e n t t i s s u e s , the c u l t u r e c o n d i t i o n s under which the c e l l s are grown can a l s o have considerable impact on t h e i r shape. Factors such as composition of the growth medium, type of su b s t r a t e on which the c e l l s are grown, and whether or not the c e l l s have grown to confluence can a f f e c t c e l l morphology (Freshney, 1987). A l l of these f a c t o r s were c a r e f u l l y c o n t r o l l e d i n the p r e p a r a t i o n of c e l l s f o r scanning w i t h the C e l l Analyzer. The type of growth medium used f o r each c e l l l i n e , once e s t a b l i s h e d , was not v a r i e d , and c e l l s were always p l a t e d at low d e n s i t i e s i n t o polystyrene t i s s u e c u l t u r e v e s s e l s . Morphology i s a l s o s i g n i f i c a n t l y a l t e r e d when c e l l s have been detached from t h e i r growth surface ( i . e . by t r y p s i n i z a t i o n ) . Some c e l l l i n e s show a f l a t t e n e d , spread out morphology when they adhere to a s u b s t r a t e , but f r e s h l y t r y p s i n i z e d c e l l s and c e l l s grown i n suspension c u l t u r e have a s p h e r i c a l shape. Figure 17 shows the appearance of V79 and CHO c e l l s w i t h i n 1-2 hours of t r y p s i n i z a t i o n , as w e l l as s e v e r a l hours l a t e r . V79 c e l l s can be seen to undergo considerable f l a t t e n i n g as they adhere to the t i s s u e c u l t u r e f l a s k . CHO c e l l s , however, do not. C e l l s t h a t are s p h e r i c a l i n shape produce l a r g e r s i g n a l peaks than do f l a t t e n e d c e l l s , and the r e f o r e are more e a s i l y detected. Thus, i n order to achieve optimum d e t e c t i o n of V79 c e l l s , they had to be scanned w i t h i n a few hours a f t e r t r y p s i n i z a t i o n . This was, however, somewhat d i f f i c u l t to accomplish i f s e v e r a l f l a s k s had to be scanned i n succession f o r a s u r v i v a l experiment. Staggered p l a t i n g of the samples was one a l t e r n a t i v e i n these circumstances, but t h i s would r e q u i r e that e i t h e r the time i n t e r v a l between t r y p s i n i z a t i o n and treatment, or the time i n t e r v a l between treatment and 6 2 a u o b d c Figure 17. V79 and CHO c e l l s at d i f f e r e n t times a f t e r t r y p s i n i z a t i o n and p l a t i n g . A l l photographs were taken using a 20X objective lens. (a) V79 c e l l s , 2 hours a f t e r p l a t i n g , (b) V79 c e l l s , 8 hours a f t e r p l a t i n g , (c) CHO c e l l s , 2 hours af t e r p l a t i n g , (d) CHO c e l l s , 8 hours a f t e r p l a t i n g . 63 p l a t i n g , be v a r i e d f o r each dose p o i n t . Varying e i t h e r of these i n t e r v a l s can a f f e c t c e l l s u r v i v a l (Reddy et a l , 1989; A l p e r , 1979). Another a l t e r n a t i v e was to t r y p s i n i z e c e l l s at staggered i n t e r v a l s f o r each dose p o i n t , but t h i s would introduce the p o s s i b i l i t y of v a r i a b i l i t y between the c e l l p o pulations used f o r the d i f f e r e n t p o i n t s . I t t h e r e f o r e appeared that a l l samples f o r a s u r v i v a l experiment would have to be prepared at the same time. Consequently, measurements were made to determine the s e v e r i t y of c e l l f l a t t e n i n g as a f u n c t i o n of in c u b a t i o n time at 37 °C. This was done by generating time-lapse records of c e l l area and bri g h t n e s s u s i n g the RSCAN program. C e l l area i n t h i s case i s the area covered by the c e l l on the surface on which i t r e s t s . C e l l b r i g h t n e s s gives an i n d i c a t i o n of how s p h e r i c a l the c e l l i s , and i s determined by measuring the volume of the s i g n a l peak over the e n t i r e c e l l and d i v i d i n g i t by the c e l l area. Figure 18 shows the time course of the changes i n these two features f o r both V79 and CHO c e l l s incubated at 37 °C. As expected, CHO c e l l s d i d not show d r a s t i c changes i n c e l l shape, ex p e r i e n c i n g an increase i n area of only about 30% between 2 hours and 10 hours a f t e r p l a t i n g . The c e l l b r i g htness changed by a s i m i l a r l y small amount. By c o n t r a s t , V79 c e l l area more than doubled between 2 hours and 7 hours p o s t - p l a t i n g . Furthermore, c e l l b r i g h t n e s s diminished c o n s i d e r a b l y over t h i s time i n t e r v a l . The continuous f l a t t e n i n g of V79 c e l l s during the f i r s t 7-8 hours a f t e r p l a t i n g s i g n i f i c a n t l y compromised the o p t i m i z a t i o n of c e l l d e t e c t i o n whenever a l a r g e number of f l a s k s had to be scanned. Furthermore, c e l l growth and d i v i s i o n occurred at a steady r a t e during t h i s i n c u b a t i o n p e r i o d f o r both V79 and CHO populations. The presence of l a r g e numbers of c e l l doublets p r i o r to scanning was und e s i r a b l e , s i n c e doublets present during treatment w i t h a damaging agent should i d e a l l y be e l i m i n a t e d from the data 64 300 cn 250-.3 200-CD <D & 100 50 (a) OOOQ O O„„«O O 0 O 0 ° o o°ooo QO 0 o oo O 0 0 o o o w o • V79 cells o CHO cells I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I n 2 3 4 5 6 7 8 9 10 ~3 50 > 0 ) £ 40 q cn 35-cn CD 5 30^ •SP 25-p i 0 . ( b ) • V79 cells o CHO cells! oo o o o o o o o o oo • • — •• ooo •• • • o ooo oo o o o o o ooo o o o oo o o o I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 2 3 4 5 6 7 8 9 10 time since plating (hr) Figu re 18. E f f e c t of in c u b a t i o n time at 37 °C on the shape of f r e s h l y t r y p s i n i z e d and p l a t e d c e l l s . (a) change i n V79 and CHO c e l l area as a f u n c t i o n of in c u b a t i o n time (1 p i x e l = 1.3 X 1.3 M m)• (b) change i n V 7 9 and CHO c e l l b r i g h t n e s s w i t h i n c u b a t i o n time. At 2 hours a f t e r p l a t i n g ( i n i t i a l p o i n t on graphs), c e l l s have a s p h e r i c a l shape. 65 set i f a s i n g l e - c e l l response i s to be measured. The i n c u b a t i o n time p r i o r to scanning t h e r e f o r e had to be short enough so that most of the doublets present during scanning were already present during treatment. V79 c e l l s , w i t h a p o p u l a t i o n doubling time of approximately 10 hours, had up to 60% doublets i n a f l a s k w i t h i n 7 hours of p l a t i n g . CHO c e l l c u l t u r e s , w i t h a s l i g h t l y longer doubling time of 11-12 hours, had 40-50% doublets w i t h i n the same time p e r i o d . R e j e c t i o n of a l l these o b j e c t s would not only r e s u l t i n a considerable r e d u c t i o n i n the number of c e l l s i n the data s e t , but t h i s data set would be composed of a p a r t i a l l y synchronized c e l l p o p u l a t i o n c o n s i s t i n g of those c e l l s that were near the beginning of the c e l l c y c l e during treatment. When the response of c e l l s to the treatment depends on t h e i r c e l l cycle, p o s i t i o n , as i s the case f o r r a d i a t i o n damage, such p a r t i a l synchrony c o u l d s i g n i f i c a n t l y a f f e c t experimental r e s u l t s . In order to prevent appreciable doublet formation before scanning, f l a s k s c o u l d be te m p o r a r i l y s t o r e d at room temperature a f t e r an i n i t i a l i n c u b a t i o n p e r i o d to a l l o w f o r c e l l attachment to the f l a s k surface (a 1-2 hour p e r i o d a t 37 °C was found to be s u f f i c i e n t f o r t h i s purpose f o r both V79 and CHO c e l l s ) . In a d d i t i o n to i n h i b i t i n g c e l l d i v i s i o n , t h i s procedure a r r e s t e d c e l l spreading i n V79 c e l l s . The p e r s i s t e n c e of a rounded morphology when V79 c e l l s were h e l d at room temperature a f t e r a 2-hour attachment p e r i o d at 37 °C i s demonstrated i n the c e l l area and bri g h t n e s s p l o t s of Figure 19. F l a t t e n i n g of the c e l l s can c l e a r l y be seen to resume once they were reint r o d u c e d to a 37 °C environment. Temporarily removing f l a s k s from the incubator a f t e r an i n i t i a l attachment p e r i o d a l s o had the advantage that a l l c e l l s used i n a given experiment could be t r e a t e d i n the same way. A l l f l a s k s c o u l d be removed from the incubator at the same time, and could a l s o be re p l a c e d at the same w 3" o H> Po r t CD X w r t rt) CD £ 3 * O c l-s cn PJ H i r t CD i-( o CD i - i CD i - i CD I CD X T3 O W P>. r t H-3 OP cn H rt) Po O CD in 3" P> 11 r*> IQ (J r( CD r t CD 3 xi CD I-i P» r t C n CD p> M l C/2 CD CU CD 3 o rr pi 3 era CD cn t o 3 OJ CD Xi CD g o 3J P) 3 OP I j j CD O" to 3" H i o n C a H cn 3" CD ^ H H- T3 O i - i Pi PJ r t p) r t CD a a* H H ' OP 3* r t 3 CD cn cn O r t O j y Pi CD r t 5 Co < o o r t P> 7? CD X} r- 1 P> n CD r t o o CD 7? CD r t H ) O P> ht r t brightness (grey levels) area (pixels) r t 3" CD n CD CTl CTl 67 time. I n the case of c e l l s u r v i v a l measurements, i t was found that such a procedure c o u l d be f o l l o w e d without s i g n i f i c a n t l y a f f e c t i n g the r e s u l t s i n any way, as long as the time spent at room temperature d i d not exceed 8-10 hours. This can be seen i n the data of Figure 20, which shows the e f f e c t of time spent at room temperature on both i r r a d i a t e d and u n i r r a d i a t e d c e l l s . The removal of f l a s k s from the incubator a f t e r a 2 hour attachment time was th e r e f o r e adopted as standard procedure f o r experiments and measurements r e q u i r i n g use of the c e l l d e t e c t i o n a l g o r i t h m , w i t h f l a s k s being returned to the incubator only when a l l scanning was completed. This ensured the presence of a p o p u l a t i o n of rounded c e l l s , w i t h only a small number of doublets present during the scanning procedure. 4.4 Development of a Focusing Algorithm 4.4.1 Determination of I d e a l Focus f o r a S i n g l e C e l l Because the c e l l d e t e c t i o n procedure searched f o r s i g n a l peaks, i d e a l focus f o r a given c e l l was defined to be the focus s e t t i n g at which the l a r g e s t peak was produced. I d e a l focus could be found by searching f o r the maximum peak as the distance of the microscope o b j e c t i v e from the stage was v a r i e d . This c o u l d be done i n t e r a c t i v e l y u s i n g the manual focus adjustment of the microscope and a g r a p h i c a l d i s p l a y of the c e l l s i g n a l , or, a l t e r n a t i v e l y , the C e l l Analyzer was capable of performing the search a u t o m a t i c a l l y . The automated search was co n s i d e r a b l y f a s t e r than the manual method, and was almost always used when accurate f o c u s i n g was re q u i r e d . The l o c a t i o n of the i d e a l focus l e v e l was c a r r i e d out by the C e l l A nalyzer through the use of a b i n a r y autofocusing procedure (Poon et a l , 1989), which operated as f o l l o w s . F i r s t , u s i n g the z - d r i v e and the d i g i t a l 0 5 10 15 20 25 30 time at room temperature (hr) Figure 20. E f f e c t of time spent at room temperature (22-25 °C) on c e l l s u r v i v a l . F r e s h l y p l a t e d c e l l s were incubated at 37 °C f o r 2 hours p r i o r to being kept at room temperature. S u r v i v a l measurements were made f o r both i r r a d i a t e d (250 k v p X-rays) and u n i r r a d i a t e d c e l l s . I r r a d i a t i o n s were performed on c e l l s i n suspension p r i o r to p l a t i n g them i n t o f l a s k s . 69 s i g n a l processor, peak height values were determined at 32 nm i n t e r v a l s from 96 pm above to 96 m^ below the c u r r e n t focus l e v e l . The two consecutive focus l e v e l s which produced the h i g h e s t o v e r a l l s i g n a l peaks were then s e l e c t e d , and a "zero" plane was d e f i n e d midway between them. A b i n a r y search f o r the i d e a l focus l e v e l would then proceed. Peak heights were determined at t e s t l e v e l s 16 /im above and 16 pm below the zero plane, and a new zero plane was defined at the p o i n t midway between the current zero plane and the t e s t plane that produced the h i g h e s t s i g n a l peak. The next t e s t planes were 8 /zm above and 8 pm below the new zero plane as the b i n a r y search continued. When a 1 /m step had been reached by c o n t i n u i n g t h i s procedure, the f i n a l zero plane defined the i d e a l focus l e v e l f o r the o b j e c t . The autofocusing procedure took approximately 5 seconds to complete. However, a d d i t i o n a l time was r e q u i r e d to l o c a t e the o b j e c t that was to be focused. Objects could be s e l e c t e d e i t h e r manually or a u t o m a t i c a l l y . An automated search i n v o l v e d scanning the f l a s k u n t i l a s i g n a l above the d e t e c t i o n t h r e s h o l d was encountered. Autofocusing would then proceed. A l t e r n a t i v e l y , o bjects could be s e l e c t e d manually u s i n g a j o y s t i c k to c o n t r o l the microscope stage movements. A c e l l would be moved to the v i c i n i t y of the center • of the l i n e a r sensor, a f t e r which i t was a u t o m a t i c a l l y centered according to i t s "center of mass" ( i n terms of grey l e v e l i n t e n s i t y ) . The center of mass was c a l c u l a t e d by the C e l l Analyzer u s i n g i n f o r m a t i o n from an automated high r e s o l u t i o n scan of a 96 (j,m by 96 pm r e g i o n around the s p e c i f i e d object l o c a t i o n . Once the s e l e c t e d o b j e c t had been centered, autofocusing would begin. 70 4.4.2 Estimation of the Acceptable Focus Range f o r C e l l Detection Because t i s s u e c u l t u r e v e s s e l s are not o p t i c a l l y f l a t , there can be c o n s i d e r a b l e v a r i a t i o n i n the i d e a l focus l e v e l as a f l a s k i s scanned. For optimum c e l l d e t e c t i o n , these v a r i a t i o n s had to be c o r r e c t e d f o r by a d j u s t i n g the focus s e t t i n g during the scanning procedure. However, because f i n d i n g the i d e a l focus l e v e l u s i n g the autofocusing procedure was a r e l a t i v e l y time-consuming process, the need f o r frequent monitoring and adjustment of the focus had the p o t e n t i a l to increase c o n s i d e r a b l y the time r e q u i r e d to scan a f l a s k . To minimize the frequency w i t h which p r e c i s e focus l e v e l s would have to be found during a scan, i t was t h e r e f o r e necessary to determine the degree of d e v i a t i o n from i d e a l focus l e v e l s that would s t i l l a l l o w adequate performance of the c e l l d e t e c t i o n procedure. Because peak height above background was the s i g n a l f e a t u r e used f o r both c e l l d e t e c t i o n and determination of the i d e a l focus l e v e l , the e f f e c t of defocusing ( i . e . d e v i a t i o n from optimum focus l e v e l s ) on t h i s f e a t u r e was measured (Spadinger et a l , 1990). This was f i r s t done u s i n g s e v e r a l d i f f e r e n t microscope o b j e c t i v e lenses, s i n c e the numerical aperture of the o b j e c t i v e a f f e c t s the s i g n a l amplitude and i t s response to defocusing. Figure 21 shows the focus dependence of V79 peak h e i g h t f o r 4.OX m a g n i f i c a t i o n o b j e c t i v e s having numerical apertures of 0.20, 0.16, and 0.13. Each curve on the graph represents the mean peak height f o r 15 c e l l s . Measurements were made w i t h 4.OX o b j e c t i v e s because i t had p r e v i o u s l y been e s t a b l i s h e d that a t o t a l image m a g n i f i c a t i o n of 10X (2.5X p r o j e c t i o n lens combined w i t h 4.OX o b j e c t i v e ) provided adequate r e s o l u t i o n f o r c e l l d e t e c t i o n and r e c o g n i t i o n purposes ( P a l c i c and J a g g i , 1986). In Figure 21a, i t can be seen that l a r g e r numerical apertures produce higher s i g n a l peaks at optimum focus. However, the amplitude of the s i g n a l peak drops o f f at a more r a p i d r a t e f o r l a r g e r numerical apertures as distance 71 100 -100 - 8 0 - 6 0 - 4 0 - 2 0 0 20 40 60 f o c u s l e v e l (jam) F i g u r e 21. E f f e c t o f d e f o c u s i n g on c e l l s i g n a l peak h e i g h t f o r 4. OX o b j e c t i v e s o f 3 d i f f e r e n t n u m e r i c a l a p e r t u r e s . I d e a l f o c u s , where peak h e i g h t i s a t a maximum, i s a t "0". Curves r e p r e s e n t a v e r a g e d d a t a from 15 c e l l s . (a) a b s o l u t e peak h e i g h t above b a c k g r o u n d l i g h t l e v e l s . (b) peak h e i g h t n o r m a l i z e d t o the maximum o c c u r r i n g a t the i d e a l f o c u s s e t t i n g . 72 from the optimum focus increases. This e f f e c t can be seen more c l e a r l y i n the normalized p l o t of Figure 21b. Thus, while the l a r g e s t peak heights would be obtained u s i n g the 4.0/0.20 o b j e c t i v e , they would a l s o be subject to the most v a r i a t i o n i f i d e a l focus l e v e l s c o u l d not be p r o p e r l y maintained. Despite the f a c t that the use of an o b j e c t i v e w i t h a smaller numerical aperture would r e s u l t i n l e s s v a r i a t i o n i n the peak height w i t h d e v i a t i o n s from optimum focus, the 4.0/0.20 o b j e c t i v e was s e l e c t e d f o r the d e t e c t i o n procedure because i t produced the l a r g e s t s i g n a l peaks. There were s e v e r a l reasons f o r making t h i s choice. F i r s t , because the c e l l d e t e c t i o n t h r e s h o l d was set at 15 grey l e v e l s above background, the f o c a l range i n which s i g n a l peaks exceeded the t h r e s h o l d was approximately the same f o r the 4.0/0.20 o b j e c t i v e as f o r the others. Furthermore, the sharper focus dependence seen w i t h higher numerical apertures was advantageous f o r l o c a t i n g the i d e a l focus s e t t i n g . F i n a l l y , because high-i n t e n s i t y s i g n a l peaks span a l a r g e r range of grey l e v e l s than low-i n t e n s i t y peaks, they c o n t a i n more infor m a t i o n . This i n c r e a s e d r e s o l u t i o n was important f o r c e l l r e c o g n i t i o n purposes. A f t e r s e l e c t i o n of the 4.0/0.20 o b j e c t i v e , the acceptable d e v i a t i o n of focus from the optimum l e v e l while a f l a s k was being scanned could be estimated. This was done by again c o n s i d e r i n g the v a r i a t i o n of V79 c e l l s i g n a l peak height as a f u n c t i o n of focus s e t t i n g . D i f f e r e n c e s i n peak height w i t h i n the c e l l p o p u l a t i o n a l s o had to be considered, however. Figure 22 shows the e f f e c t of defocusing on peak height f o r a p o p u l a t i o n of 76 V79 c e l l s . The f r a c t i o n of c e l l s w i t h peak heights below the 15 grey l e v e l d e t e c t i o n t h r e s h o l d i s a l s o shown. From t h i s p l o t , i t i s evident t h a t e s s e n t i a l l y a l l c e l l s should be detected i f the focus s e t t i n g i s kept w i t h i n 60 fim below (microscope o b j e c t i v e too f a r from sample) and 30 /im Figure 22. E f f e c t of defocusing on c e l l s i g n a l peak height r e l a t i v e to background l i g h t l e v e l s (4.0/0.20 o b j e c t i v e ) . Both the mean (data p o i n t s ) and the standard d e v i a t i o n ( e r r o r bars) are shown f o r a p o p u l a t i o n of 76 c e l l s . The f r a c t i o n of c e l l s w i t h peaks below the d e t e c t i o n t h r e s h o l d of 15 grey l e v e l s above background i s al s o p l o t t e d ( s c a l e on right-hand a x i s ) . 74 above (microscope o b j e c t i v e too c l o s e to sample) optimum focus. Thus, the goal f o r the design of a foc u s i n g a l g o r i t h m was to maintain the focus s e t t i n g comfortably w i t h i n these l i m i t s . 4.4.3 Measurement of Flask Surface Shape The shapes of the bottom p l a t i n g surfaces of s e v e r a l Nunclon 25 cm 2 t i s s u e c u l t u r e f l a s k s were measured i n order to e s t a b l i s h the degree of focus c o r r e c t i o n that would be r e q u i r e d when scanning f l a s k s of t h i s type. I t was a l s o necessary to determine whether the i d e a l focus c o u l d vary by more than the all o w a b l e amount across the width of any of the 2.25 mm wide bands scanned by one sweep of the l i n e a r sensor. Focus adjustments could only be made between successive l i n e scans ( i . e . along the len g t h of a band) or between successive scanning bands. Optimum focus l e v e l s were measured by f o c u s i n g on c e l l s at s e l e c t e d p o i n t s i n the 4.0 X 2.5 cm scanning area i n each of s e v e r a l f l a s k s . C e l l s were p l a t e d at r e l a t i v e l y h i g h d e n s i t i e s (8-10 cells/mm 2) i n these f l a s k s so t h a t measurements from more than one c e l l c o u l d be used to define the i d e a l focus l e v e l at the same approximate " p o i n t " . Measurements were taken at 9 " p o i n t s " along the c e n t e r l i n e of each band, y i e l d i n g a t o t a l of 99 readings f o r the scanning area. Each of the 99 readings c o n s i s t e d of the mean optimum focus l e v e l f o r 2-3 c e l l s i n c l o s e p r o x i m i t y to each other. In a d d i t i o n to e s t a b l i s h i n g the o v e r a l l shape of the f l a s k s u rface, the focus measurements allowed the i d e n t i f i c a t i o n of r e l a t i v e l y f l a t regions w i t h i n the scanning area and provided i n f o r m a t i o n as to whether a c h a r a c t e r i s t i c surface shape could be i d e n t i f i e d f o r t h i s type of f l a s k . I f a l l Nunclon 25 cm 2 f l a s k s were to possess a s i m i l a r shape, a foc u s i n g a l g o r i t h m c o u l d be designed to e x p l o i t t h i s property. A f t e r making measurements on s e v e r a l f l a s k s , a c h a r a c t e r i s t i c surface shape was indeed 75 evident. This shape i s i l l u s t r a t e d by data from a t y p i c a l f l a s k , shown i n Figure 23. The optimum focus l e v e l along each band f l u c t u a t e d by no more than about ±20 /im from the mean value f o r that band. Establishment of an appropriate focus s e t t i n g f o r a band t h e r e f o r e meant that f u r t h e r focus adjustments along i t s length d i d not have to be made. Adjustments s t i l l had to be made between bands, however, si n c e mean focus l e v e l s f o r i n d i v i d u a l bands v a r i e d by as much as 100 /im between the edges and the center of the scanning area. F o r t u n a t e l y , t h i s v a r i a t i o n was not enough to cause excessive focus d e v i a t i o n s across the width of any of the scanning bands. 4 . 4 . 4 * A Semi-Automated Focusing Procedure Because of the c h a r a c t e r i s t i c shape of the Nunclon f l a s k s u r f a c e , i t was p o s s i b l e to design a r e l a t i v e l y simple, "semi-automated" f o c u s i n g procedure f o r these v e s s e l s . The procedure i s r e f e r r e d to as semi-automated because the c e l l s used f o r focus measurements are s e l e c t e d manually r a t h e r than a u t o m a t i c a l l y . The determination of the optimum focus l e v e l f o r each c e l l i s , however, performed a u t o m a t i c a l l y u s i n g the b i n a r y autofocusing procedure ( S e c t i o n 4.4.1). Aside from the r e q u i r e d presence of the experimenter to l o c a t e and i d e n t i f y c e l l s , manual s e l e c t i o n had s e v e r a l advantages over automated s e l e c t i o n . Because the experimenter had a l a r g e r f i e l d of view through the microscope than the l i n e a r sensor, he or she could randomly s e l e c t c e l l s more r a p i d l y than the C e l l Analyzer. In a d d i t i o n , the experimenter could d i s t i n g u i s h c e l l s from d e b r i s w i t h greater c e r t a i n t y , e s p e c i a l l y i f the focus l e v e l at which the search was made was f a r from i d e a l . I t was very important t h a t i d e a l focus l e v e l s were determined from c e l l s r a t h e r than d e b r i s , s i n c e d e b r i s s i g n a l s were f r e q u e n t l y maximized at focus l e v e l s f a r 76 10 crrr scanning area consisting of 11 adjacent scanning bands band 1 band 3 band 5 band 7 band 9 band 11 0 5 10 15 20 25 30 35 40 x coordinate (mm) Figure 23. The shape of the p l a t i n g surface i n a t y p i c a l Nunclon 25 cm2 t i s s u e c u l t u r e f l a s k . V a r i a t i o n s i n the i d e a l focus l e v e l , as determined by maximization of the c e l l s i g n a l peak height, are shown f o r s e l e c t e d 2.25 mm wide scanning bands i n the re g i o n of the f l a s k t y p i c a l l y scanned by the CSCAN and CELREC programs. Focus measurements on 2-3 c e l l s at the same approximate l o c a t i o n i n the f l a s k were averaged to o b t a i n the data p o i n t s . The "0" i d e a l focus l e v e l f o r the p l o t s i s a r b i t r a r y . 77 from the optimum focus f o r nearby c e l l s . The experimenter c o u l d a l s o immediately a s c e r t a i n whether the c u r r e n t focus d e v i a t e d c o n s i d e r a b l y from the i d e a l , and could, i f necessary, make a coarse focus adjustment p r i o r to implementation of the autofocusing procedure. Since i t had been found that focus adjustments were r e q u i r e d only between bands when scanning the Nunclon f l a s k s , optimum focus l e v e l s had to be e s t a b l i s h e d f o r , at most, each of the 11 scanning bands. Furthermore, the v a r i a t i o n i n optimum focus l e v e l s between scanning bands was e s s e n t i a l l y p a r a b o l i c , a l l o w i n g easy i n t e r p o l a t i o n between p o i n t s . Indeed, i t was found t h a t as few as three a p p r o p r i a t e l y spaced p o i n t s c o u l d be used to f i t a q u a d r a t i c curve that provided a good approximation to the focus v a r i a t i o n across the e n t i r e 11 band width of the scanning area. A three p o i n t f i t to the data of Figure 23 i s shown i n Figure 24. This f i t , which used the mean optimum focus l e v e l s e s t a b l i s h e d f o r bands 1, 6, and 11, d e v i a t e d from the measured values f o r the 11 bands by no more than 5-10 um. Thus, i d e a l focus l e v e l s had to be e s t a b l i s h e d only f o r the center band and the two outside bands i n order to c h a r a c t e r i z e the shape of the f l a s k surface w i t h i n the scanning area. This allowed a r e l a t i v e l y f a s t and simple f o c u s i n g procedure to be designed f o r the Nunclon f l a s k s . A f l o w c h a r t of the semi-automated f o c u s i n g procedure i s shown i n Figure 25. The f i r s t step i n v o l v e d the determination of the optimum focus l e v e l f o r s e l e c t e d bands i n the scanning area. Focus l e v e l s had to be e s t a b l i s h e d f o r at l e a s t 3 bands (bands 1, 6, and 11), but c o u l d a l s o be determined f o r more bands i f d e s i r e d . The appropriate focus s e t t i n g f o r a band was determined by averaging the optimum focus l e v e l s f o r one or more c e l l s l o c a t e d near the band c e n t e r l i n e . As mentioned p r e v i o u s l y , these c e l l s would be l o c a t e d manually by the experimenter, but t h e i r i d e a l focus l e v e l would be found w i t h the b i n a r y autofocusing procedure. The 78 0 5 10 15 20 25 y coordinate (mm) Figure 24. Mean i d e a l f o c u s l e v e l s f o r the 11 s c a n n i n g bands i n a t y p i c a l N u n c l o n 25 cm 2 f l a s k (same f l a s k as F i g u r e 23) . E r r o r b a r s a r e s t a n d a r d d e v i a t i o n s . The c u r v e used t o i n t e r p o l a t e between d a t a p o i n t s was o b t a i n e d by f i t t i n g a q u a d r a t i c e q u a t i o n t o o n l y 3 o f the d a t a p o i n t s . The p o i n t s u s e d a r e i n d i c a t e d by the open c i r c l e s . 79 user selects to focus in 3, 6, or 11 bands drive stage to center of band 1 stage under joystick control (user si ilects a c e l l ) stage under computer control drive stage to center of next band center and autofocus selected object display z -value for focused object input user option: (1) accept current focus value and take another reading in this band (2) accept current focus value and move to next band (3) reject current focus value and take another reading in this band (4) reject current focus value and move to next band yes yes yes average a l l focus values accepted for current band and store yes f i t quadratic curve to averaged focus values for selected bands calculate focus levels for the 11 scanning bands from the quadratic f i t Figure 25. Flow chart of the semi-automated fo c u s i n g r o u t i n e f o r Nunclon f l a s k s . 80 microscope stage was a u t o m a t i c a l l y d r i v e n to the center of each s e l e c t e d band to ensure that focus readings were obtained i n appropriate regions of the f l a s k . Once a l l necessary focus measurements had been made, a quadr a t i c l e a s t squares f i t to the p o i n t s was c a l c u l a t e d . The focus s e t t i n g s to be used f o r each of the 11 bands i n the scanning area were determined from t h i s f i t , and adjustments were then made a u t o m a t i c a l l y between successive bands as the f l a s k was scanned. 4.4.5 F u l l y Automated Focusing Procedures F u l l y automated f o c u s i n g procedures that would not r e q u i r e the presence of the experimenter f o r c e l l s e l e c t i o n purposes were a l s o proposed and implemented. One of these procedures fun c t i o n e d i n a s i m i l a r way to the semi-automated procedure described above, except t h a t the C e l l Analyzer would a u t o m a t i c a l l y l o c a t e the objects that were used to e s t a b l i s h the optimum focus l e v e l s at the s e l e c t e d band centers. A second f u l l y automated f o c u s i n g procedure that was implemented would focus on objects encountered at s e l e c t e d i n t e r v a l s w i t h i n each band as i t was scanned. This procedure c o u l d be used f o r t i s s u e c u l t u r e v e s s e l s that d i d not have a known c h a r a c t e r i s t i c shape, si n c e i t d i d not r e l y on predetermined i n f o r m a t i o n about the v e s s e l surface. I n c o r r e c t values f o r the optimum focus l e v e l were, however, o f t e n obtained when objects on which to focus were s e l e c t e d a u t o m a t i c a l l y . These e r r o r s arose from two primary sources. The f i r s t of these was the s e l e c t i o n of de b r i s r a t h e r than c e l l s , which c o u l d sometimes l e a d to focus s e t t i n g s that were i n c o r r e c t by as much as 100 /xm. The search procedure, which would seek a peak above the d e t e c t i o n t h r e s h o l d without regard to the form of the s i g n a l i t s e l f , c ould not d i s t i n g u i s h between c e l l s and d e b r i s . 81 The second major source of i n c o r r e c t s e t t i n g s occurred when small peaks caused by d i f f r a c t i o n at the c e l l edges (the " b r i g h t rim") were hig h enough to exceed the d e t e c t i o n t h r e s h o l d . This would commonly occur when the focus l e v e l at which the search was made was 60-120 //m above the i d e a l focus l e v e l f o r the curr e n t l o c a t i o n i n the t i s s u e c u l t u r e v e s s e l . In t h i s focus range, the c e l l s i g n a l had d e t e r i o r a t e d i n such a way th a t the c e n t r a l peak had disappeared while the amplitude of the b r i g h t rim increased, as shown i n Figure 26. A s i m i l a r d e t e r i o r a t i o n of the s i g n a l d i d not occur at focus l e v e l s below the i d e a l , where the form of the s i g n a l was e s s e n t i a l l y preserved as the amplitude of the peak decreased. Because the c e n t r a l peak was small or nonexistent near the c e l l edge, a b i n a r y search f o r the maximum peak i n t h i s r e g i o n ended up maximizing the peak produced by the b r i g h t rim, r e s u l t i n g i n " i d e a l " focus l e v e l s t h a t were approximately 80 nm higher than the true i d e a l focus f o r the c e l l . This can be seen from p l o t s of the maximum s i g n a l i n t e n s i t y as a f u n c t i o n of focus s e t t i n g at v a r i o u s distances from the c e l l center (Figure 27, compare a l s o to Figure 22). C l e a r l y , a l a r g e p r o p o r t i o n of the c e l l p o p u l a t i o n produced detectable s i g n a l s from the b r i g h t rim at 60-120 /im above the i d e a l focus l e v e l . One simple method used to reduce e r r o r s caused by f o c u s i n g on c e l l edges was to autofocus on each c e l l twice. The f i r s t f o c u s i n g would be c a r r i e d out wherever a peak was detected, and the second would be performed 4 nm f u r t h e r along the scanning (x-) d i r e c t i o n ( r e f e r to Figure 16). The hig h e s t peak height value obtained f o r the two readings would then be used to i d e n t i f y the c o r r e c t focus l e v e l . This method worked reasonably w e l l , s i n c e the maximum peak at the c e l l center was c o n s i d e r a b l y higher than any peaks generated by the b r i g h t rims. However, because b r i g h t rims could be detected as f a r as 8 fim from the c e l l center, a 4 /xm step was not always 82 C/2 "HJ > CD CD • i—i CD 2 0 0 180 160 140 120 100 80 60 40-1 20 1 8 0 -1 6 0 -140 120 100 8 0 -6 0 -4 0 -20 (a) -80 um (b) -60 um (c) -30 um (d) 0 um (e) +20 um (f) +30 um (g) +50 um (h) +80 um rrTJTTl T.J HT T | TI 1 T J11 Tl J111 I [ .1 I 1 F j I I 1 I ] 1'M 1 ['IT! I | I III [ I! 11 | II ! 11 I I I 11TIII | II W [ I III | I III | III F 11 I I I [ I I I I | 1 I I lTmTJTT T T1I 1 11J I IT 1 111 I 1|I 1 I I 20 25 30 35 40 4 5 20 25 30 35 40 4 5 20 25 30 35 4 0 4 5 20 25 30 35 40 4 5 pixel n o . Figure 26. E f f e c t of defocusing on the c e l l s i g n a l . A s i g n a l from a V79 c e l l i s shown at va r i o u s stages of defocusing. (a) microscope o b j e c t i v e 80 /im below i d e a l focus. (b) 60 /im below i d e a l focus. (c) 30 /*m below i d e a l focus. (d) at i d e a l focus. (e) 20 m^ above i d e a l focus. ( f ) 30 m^ above i d e a l focus. (g) 50 m^ above i d e a l focus. (h) 80 fim above i d e a l focus. 83 > S-i GO CO 30 20-i 15^  10 5 0 (a) \ max intensity fraction below / threshold -100 -50 • i i i 0 30 50 T — | i i i—r 100 1.0 0.8 0.6 0.4 I-0.2 0.0 150 fraction below \ threshold -100 -50 focus level (|im) o t o o 1.0 ^  o o Figure 27. E f f e c t of defocusing on c e l l s i g n a l i n t e n s i t y near the c e l l edges. I n t e n s i t y i s measured r e l a t i v e to background l i g h t l e v e l s (4.0/0.20 o b j e c t i v e ) . The mean (data p o i n t s ) and the standard d e v i a t i o n ( e r r o r bars) are shown f o r a p o p u l a t i o n of 48 c e l l s . The f r a c t i o n of c e l l s w i t h maximum i n t e n s i t y peaks below the d e t e c t i o n threshold of 15 grey l e v e l s above background i s a l s o p l o t t e d ( s c a l e on right-hand axes). (a) l i n e scans taken 4 /*m from the c e l l center. (b) l i n e scans taken 8 /im from the c e l l center. I d e a l ("0") focus l e v e l was determined by maximizing the s i g n a l peak height at the c e l l center. 84 s u f f i c i e n t to reach the approximate center of the c e l l . Use of a l a r g e r step d i d not solve t h i s problem, s i n c e the centers of s m a l l e r c e l l s c ould then be completely skipped over. Focus l e v e l s from three successive focusings spaced at 4 um i n t e r v a l s would have had to be compared i n order to assure that the c e l l center was always found, but t h i s would add c o n s i d e r a b l y to the time r e q u i r e d to e s t a b l i s h the c o r r e c t focus s e t t i n g f o r a c e l l . Because of the problems w i t h proper i d e n t i f i c a t i o n and C e n t e r i n g of c e l l s i n the f u l l y automated f o c u s i n g r o u t i n e s , they were only used o c c a s i o n a l l y . C o r r e c t i o n of the problems w i t h these r o u t i n e s would r e q u i r e a more s o p h i s t i c a t e d procedure than simple a p p l i c a t i o n of the autofocusing procedure to the f i r s t s i g n a l peak encountered. Centering of the detected o b j e c t p r i o r to searching f o r the i d e a l focus, as was done f o r the manually s e l e c t e d o bjects i n the semi-automated f o c u s i n g r o u t i n e , would prevent unnecessary searches on c e l l edges. Centering an o b j e c t would take c o n s i d e r a b l y l e s s time than having to perform 2 or 3 autofocusings per ob j e c t . Once an o b j e c t was centered and focused, i t s s i g n a l f e a t u r e s would s t i l l have to be analyzed to ensure that i t was a c e l l and not a piece of d e b r i s . 4. 5 S e l e c t i o n of Step S i z e I f a l l c e l l s i n the scanning area were to be found by the d e t e c t i o n procedure, the maximum allowable step s i z e between successive l i n e scans was r e s t r i c t e d by the s i z e of the c e l l s i n the p o p u l a t i o n of i n t e r e s t . In p r i n c i p l e , a s i n g l e l i n e scan taken near the center of a c e l l was s u f f i c i e n t f o r i t s d e t e c t i o n . Furthermore, a l i n e scan taken at approximately 1 /im s p a t i a l r e s o l u t i o n and 256 grey l e v e l photometric 85 r e s o l u t i o n would l i k e l y provide s u f f i c i e n t i n f o r m a t i o n to a u t o m a t i c a l l y d i s t i n g u i s h c e l l s from d e b r i s ( P a l c i c and J a g g i , 1986). These requirements were met by the C e l l Analyzer i f 10X m a g n i f i c a t i o n was used. Both c e l l d e t e c t i o n and r e c o g n i t i o n could t h e r e f o r e be c a r r i e d out on the b a s i s of a s i n g l e " h i t " per c e l l . I t was advantageous to detect each c e l l j u s t once, not only f o r maximization of the scanning speed, but a l s o f o r the e x c l u s i o n of near neighbours. A near-neighbour e x c l u s i o n procedure c o u l d be made much simpler i f each c e l l was detected only once, si n c e t h i s allowed c e l l l o c a t i o n s to be unambiguously defined. For the e x c l u s i o n procedure to f u n c t i o n p r o p e r l y , however, a l l c e l l s i n the scanning area had to be detected. The i d e a l step s i z e was t h e r e f o r e one t h a t was s m a l l enough to ensure that a l l c e l l s were detected, but a l s o l a r g e enough so that each c e l l would be detected only once (Spadinger et a l , 1989). As p r e v i o u s l y described ( S e c t i o n 4.2), the d e t e c t i o n a l g o r i t h m used by the C e l l Analyzer recorded only one " h i t " f o r every four image l i n e s on which a s i g n a l was detected at the same approximate p i x e l l o c a t i o n . I t was t h e r e f o r e p o s s i b l e to f i n d a step s i z e that would cause the m a j o r i t y of c e l l s i n the scanning area to be " h i t " only once, even i f there was c o n s i d e r a b l e v a r i a t i o n i n c e l l s i z e w i t h i n the p o p u l a t i o n . To f i n d the appropriate step s i z e f o r a given c e l l l i n e , the range of c e l l s i z e s i n a t y p i c a l p o p u l a t i o n was f i r s t determined. This was done by measuring the d i s t r i b u t i o n of s i g n a l peak widths ( i n terms of the number of p i x e l s above the d e t e c t i o n t h r e s h o l d i n the s i g n a l peak) f o r l i n e scans across s e v e r a l randomly s e l e c t e d c e l l s . For a p o p u l a t i o n of rounded c e l l s , t h i s f e a t u r e measures the diameter of the r e g i o n of each c e l l t hat i s " v i s i b l e " to the d e t e c t i o n algorithm. Each c e l l was i n d i v i d u a l l y focused and centered p r i o r to performing the measurement. The d i s t r i b u t i o n of peak widths f o r both V79 and CHO c e l l s i s shown i n Figure 28. Considering that 0.40 a 0.35 •! | 0.30^  .2 0.25^  o OS 0.20-i 0.15-| 0.10-3 §• 0.05^  0.00 • V79 cells • CHO cells 2 4 6 8 10 12 14 peak width at threshold (pixels) Figure 28. V79 and CHO c e l l s i g n a l peak widths at the d e t e c t i o n (15 grey l e v e l s above background). 1 p i x e l = 1.3 /xm. 87 the e f f e c t i v e p i x e l width was 1.3 /xm, peak widths f o r V79 c e l l s ranged from approximately 5 to 13 /xm, w i t h a mean and standard d e v i a t i o n of 7.7 /xm and 1.6 /xm, r e s p e c t i v e l y . Peak widths f o r CHO c e l l s were l a r g e r , v a r y i n g from 5 /xm to 17 /xm, w i t h a mean of 10.5 /xm and a standard d e v i a t i o n of 2.0 /xm. I f a l l c e l l s i n the scanning area of a f l a s k were to be " h i t " at l e a s t once during an automated search, the s i z e of the steps taken between scans had to be s l i g h t l y smaller than the diameter of the " v i s i b l e " r e g ion f o r the s m a l l e s t c e l l s i n the p o p u l a t i o n . Based on t h i s c r i t e r i o n , the step s i z e f o r both V79 and CHO c e l l s was estimated to be somewhat l e s s than 5 /xm. Since the microscope stage moved i n 1 /xm increments, the l a r g e s t a v a i l a b l e step s i z e that would meet t h i s c r i t e r i o n was 4 /xm long. In order to v e r i f y the expected performance of the 4 /xm step s i z e i n comparison to that of other step s i z e s , the a c t u a l number of " h i t s " recorded per V79 c e l l by the d e t e c t i o n a l g o r i t h m was measured f o r steps ranging from 2 /xm to 8 /xm. The number of undetected c e l l s was determined u s i n g the CELREC r e v i s i t i n g mode th a t simulated the scanning procedure i n such a way that undetected objects could be v i s u a l l y i d e n t i f i e d . Only a small p a r t of the f l a s k was scanned so as to minimize the e f f e c t of improper focus on the r e s u l t s . The r e g i o n used was band 6 of the scanning area, s i n c e t h i s was i n the f l a t t e s t p a r t of the Nunclon f l a s k ( c f . Figures 23 and 24). This band was scanned and r e v i s i t e d once f o r each of the step s i z e s t e s t e d . A s i n g l e focus s e t t i n g was used f o r a l l the scans. This s e t t i n g was e s t a b l i s h e d by averaging the i d e a l focus l e v e l s measured f o r 10 or more c e l l s i n the s e l e c t e d band. The r e s u l t s f o r the number of h i t s per c e l l as a f u n c t i o n of step s i z e are shown i n Figure 29. From these data, i t i s evident that a step s i z e of e i t h e r 3 or 4 /xm would provide the d e s i r e d s i n g l e " h i t " per c e l l . Step s i z e s below 3 /xm r e s u l t e d i n a r e l a t i v e l y l a r g e p r o p o r t i o n of m u l t i p l e 88 O O a o 0.6 A OA A o 0 h i t s - • - 1 h i t >1 h i t « 0.2-1 0,0-9* "i-rTfr-rTf o r r i r n I 1 1 1 1 I 1 6 7 step s ize (]im) .0 ~i—r 8 F i g u r e 29. Dependence o f the number o f " h i t s " p e r V79 c e l l on the s t e p s i z e between s u c c e s s i v e l i n e s c a n s i n the c e l l d e t e c t i o n p r o c e d u r e . 89 h i t s per c e l l , w h i l e the number of undetected c e l l s rose s h a r p l y f o r step s i z e s g r e a t e r than 4 /im, as was expected from the d i s t r i b u t i o n of peak widths f o r V79 c e l l s . Thus, the step s i z e s e l e c t e d on the b a s i s of the peak width d i s t r i b u t i o n f o r V79 c e l l s performed as expected, producing a s i n g l e h i t per c e l l f o r a l l but a small p r o p o r t i o n of the p o p u l a t i o n . A s l i g h t l y s m a l l e r step s i z e (3 /im) was a l s o found to produce the d e s i r e d r e s u l t . This was most l i k e l y due to the f a c t that the l a r g e s t V79 c e l l s had a peak width of 12-13 /im, which was b a r e l y l a r g e enough to produce m u l t i p l e h i t s per c e l l when a 3 /im step was used. Nevertheless, the f i n a l step s i z e s e l e c t e d f o r V79 c e l l d e t e c t i o n purposes was 4 /im, s i n c e t h i s allowed more r a p i d scanning to be c a r r i e d out. A step s i z e of 4 /im was a l s o used f o r CHO c e l l s , s i n c e the s m a l l e s t c e l l s i n the p o p u l a t i o n had peak widths comparable to those of the s m a l l e s t V79 c e l l s . Although not e x t e n s i v e l y t e s t e d f o r CHO c e l l s , t h i s step s i z e was found to perform w e l l i n p r a c t i c e . A 3 /im step would, however, l i k e l y have r e s u l t e d i n a higher incidence of m u l t i p l e h i t s f o r t h i s c e l l l i n e than f o r V79 c e l l s , s ince the l a r g e s t CHO c e l l s generated c o n s i d e r a b l y wider peaks (17 /tm) than the l a r g e s t V79 c e l l s . With a 4 /im step s i z e , the 4 cm X 2.5 cm scanning area of a f l a s k c o u l d be searched i n j u s t under 13 minutes. A l l o w i n g f o r 5-6 minutes to change f l a s k s and c a r r y out the necessary focus measurements f o r the semi-automated f o c u s i n g r o u t i n e , approximately 24 f l a s k s c o u l d be scanned w i t h i n the 8-10 hour p e r i o d during which c e l l s d i d not appear to be s i g n i f i c a n t l y a f f e c t e d by being kept at room temperature. A s u r v i v a l experiment i n v o l v i n g 6 dose p o i n t s f o r each of two s u r v i v a l curves c o u l d t h e r e f o r e be performed comfortably even i f there were 2 f l a s k s per dose p o i n t . However, f o r l a r g e r numbers of dose p o i n t s or more than 2 s u r v i v a l curves per 90 experiment, only a s i n g l e f l a s k per dose p o i n t c o u l d be scanned w i t h i n the a l l o t t e d time l i m i t . 4 . 6 Focus Dependence of C e l l D e t e c tion Using S e l e c t e d Step S i z e Once the appropriate step s i z e had been s e l e c t e d f o r V79 c e l l s , the e f f e c t of defocusing on c e l l d e t e c t i o n was measured d i r e c t l y . The method employed was the same as that used to o b t a i n the data of Figure 29, except that the focus l e v e l r a t h e r than the step s i z e was v a r i e d . The r e s u l t s of these measurements are shown i n Figure 30. The "0" ( i d e a l ) focus l e v e l of the p l o t corresponds to the average i d e a l focus s e t t i n g f o r approximately 10 c e l l s i n the area scanned. Undetected c e l l s were i d e n t i f i e d e i t h e r by u s i n g the manual r e v i s i t i n g mode that simulated the scanning procedure, or by comparing the t o t a l number of c e l l s detected i n the scanning area at d i f f e r e n t focus l e v e l s . I t was necessary to use the l a t t e r method whenever l a r g e numbers of c e l l s were missed, but i t was a l s o used at intermediate focus l e v e l s f o r some t r i a l s as i t was l e s s l a b o u r - i n t e n s i v e and gave comparable r e s u l t s to the former method. N i n e t y - f i v e percent or more of the c e l l s i n the scanning area were detected a t focus l e v e l s f a l l i n g w i t h i n approximately -40 /im to +30 pm of i d e a l focus. This range was s l i g h t l y s m a ller than the acceptable f o c a l range of -60 pm to +30 /um p r e d i c t e d from peak height measurements on i n d i v i d u a l l y focused c e l l s (Figure 22). Nevertheless, the curve c a l c u l a t e d from the peak height measurements approximated the p r o p o r t i o n of undetected c e l l s very w e l l i n regions outside the -70 to -40 nm range. The discrepancy i n the -70 to -40 fim r e g i o n may be p a r t i a l l y a t t r i b u t a b l e to the f a c t that the i d e a l focus l e v e l used to determine the data p o i n t s of Figure 30 was a mean value f o r the e n t i r e band being scanned, w h i l e the 91 -100 -80 -60 -40 -20 0 20 40 60 f o c u s l e v e l ( u m ) Figure 30. Effect of defocusing on V79 c e l l detection using a 4 step size. Data were collected from the central scanning band of 3 different f l a s k s . 92 data used to c a l c u l a t e the curve were obtained from i n d i v i d u a l l y focused c e l l s . Since i d e a l focus l e v e l s i n the band v a r i e d by ±10-20 /xm, c e l l d e t e c t i o n near the l i m i t s of the acceptable focus range c o u l d have been a f f e c t e d . I n a d d i t i o n , the e f f e c t i v e diameter of the c e l l s (as measured by the peak width at the d e t e c t i o n threshold) was s h a r p l y reduced beyond approximately 40 /xm below and 30 /xm above i d e a l focus. At these focus l e v e l s , t h e r e f o r e , the step s i z e used by the scanning procedure was too la r g e to detect a l l the c e l l s . This may e x p l a i n the r e d u c t i o n i n the f r a c t i o n of c e l l s detected i n the -60 to -40 /xm range. 93 5. DEVELOPMENT OF A RECOGNITION ALGORITHM FOR V79 CELLS 5.1 Recognition C r i t e r i a Because the d e t e c t i o n a l g o r i t h m used by the C e l l Analyzer searched only f o r s i g n a l peaks r i s i n g above a s p e c i f i e d i n t e n s i t y t h r e s h o l d , i t detected c e r t a i n kinds of d e b r i s i n a d d i t i o n to the c e l l s i n the f l a s k . Types of d e b r i s t h a t could produce detectable s i g n a l s i n c l u d e d f i b r e s , dust, p r e c i p i t a t e s from the growth medium, g l a s s and p l a s t i c shards, and l y s e d c e l l s . Imperfections i n the p l a s t i c of the t i s s u e c u l t u r e v e s s e l were a l s o sometimes detected. The amount of d e b r i s detected i n a f l a s k v a r i e d between 10-40% of a l l the objects found at the c e l l d e n s i t i e s used f o r most experiments (1-2 cells/mm 2). An automated r e c o g n i t i o n a l g o r i t h m had to c o r r e c t l y i d e n t i f y n o n - c e l l objects w i t h s u f f i c i e n t accuracy to reduce t h i s p r o p o r t i o n to acceptable l e v e l s . The goal was to achieve an accuracy comparable to that of an experienced human observer, p a r t i c u l a r l y i n terms of f a l s e p o s i t i v e e r r o r , which i s c r u c i a l i n s u r v i v a l s t u d i e s . The d e f i n i t i o n of f a l s e p o s i t i v e e r r o r , along w i t h other measures of the performance of a c l a s s i f i e r , i s shown i n Figure 31. The f a l s e p o s i t i v e e r r o r corresponds to the p r o p o r t i o n of objects that has been c l a s s i f i e d as " c e l l " but i s a c t u a l l y d e b r i s . N o n - c e l l objects mistakenly i n c l u d e d i n a data set were not always i d e n t i f i a b l e during subsequent measurements and could, i f present i n s i g n i f i c a n t numbers, l e a d to a p p r e c i a b l e e r r o r s i n the r e s u l t s of s u r v i v a l experiments. This would occur because n o n - c e l l s would not be e a s i l y d i s t i n g u i s h a b l e from non-survivors when parent c e l l l o c a t i o n s were being checked f o r colony formation. Debris l o c a t i o n s would be c l a s s i f i e d as non-survivors, thereby i n f l a t i n g the number of objects i n t h i s category. 94 Confusion Matrix: Manual C l a s s i f i c a t i o n Recognition Algorithm c e l l non-cell t o t a l c e l l a b a + b non-cell c d c + d t o t a l a + c b + d a + b + c + d a x 100 % c e l l s correct = (a + 6) d x 100 % non-cells correct = (c + d) Defining cells as "positives": % f a l s e p o s i t i v e % f a l s e negative = % t o t a l error = c x 100 (a + c) & x 100 (6 + d) (b + c) x 100 (a + b + c + d) Figure 31. Confusion matrix and formulae f or the evaluation of recognition algorithm performance. The quantities a, b, c, and d are the numbers of objects i n the designated categories. 95 In c o n t r a s t to the f a l s e p o s i t i v e e r r o r , the f a l s e negative e r r o r , which r e s u l t s from c e l l s being m i s c l a s s i f i e d as d e b r i s , would not d i r e c t l y a l t e r the outcome of an experiment. I t s primary e f f e c t would be to reduce the p o t e n t i a l s i z e of the data set. This e r r o r t h e r e f o r e d i d not need to be as s t r i c t l y c o n t r o l l e d as the f a l s e p o s i t i v e e r r o r , although i t could not be so l a r g e that a s i g n i f i c a n t number of c e l l s was l o s t from the data s e t . To d e f i n e a reasonable upper l i m i t f o r the a l l o w a b l e f a l s e p o s i t i v e e r r o r r a t e generated by an automated r e c o g n i t i o n a l g o r i t h m , the "worst case" expected to a r i s e when using manual c l a s s i f i c a t i o n was approximated. An experienced observer can c l a s s i f y 97-100% of c e l l s and 94-100% of debris c o r r e c t l y on the day of p l a t i n g (the performance of the human observer was measurable s i n c e objects could be u n e q u i v o c a l l y i d e n t i f i e d as c e l l s by i n c u b a t i n g them at 37 °C f o r 2-3 days and checking whether or not they had p r o l i f e r a t e d ) . The "worst case" would be a f l a s k i n which 97% of c e l l s and 94% of d e b r i s were c o r r e c t l y c l a s s i f i e d . Furthermore, s i n c e f a l s e p o s i t i v e e r r o r r a t e s were a f f e c t e d by the p r o p o r t i o n of d e b r i s , 40% of the detected o b j e c t s would be d e b r i s . Under these c o n d i t i o n s , a 4% f a l s e p o s i t i v e e r r o r would be generated. Thus, the goal was to design an automated r e c o g n i t i o n a l g o r i t h m t h a t produced f a l s e p o s i t i v e e r r o r r a t e s of l e s s than 5%. Since V79 c e l l s were to be used f o r most experiments, the a l g o r i t h m was developed s p e c i f i c a l l y f o r t h i s c e l l l i n e . 96 5.2 Data A c q u i s i t i o n The development and e v a l u a t i o n of an a l g o r i t h m to c l a s s i f y o b jects detected by the C e l l Analyzer r e q u i r e d the c o l l e c t i o n of l a r g e sample data s e t s . These samples had to be r e p r e s e n t a t i v e of the q u a l i t y of data obtainable under experimental c o n d i t i o n s . Thus, the same procedure was used to c o l l e c t sample data sets as was used f o r the l o c a t i o n of c e l l s d u r i n g the course of an experiment. This procedure was described p r e v i o u s l y i n Chapter 4. In summary, c e l l s were p l a t e d at low d e n s i t i e s (1-2 cells/mm 2) i n t o the 25 cm 2 Nunclon t i s s u e c u l t u r e f l a s k s and incubated f o r 2 hours p r i o r to scanning. Flasks were scanned u s i n g a 4 /xm step s i z e . Focus was maintained during the scans by the semi-automated f o c u s i n g r o u t i n e of S e c t i o n 4.4.4. Image data from a l l objects detected i n a number of d i f f e r e n t f l a s k s were recorded and s t o r e d so that the c h a r a c t e r i s t i c s of c e l l and n o n - c e l l s i g n a l s c o u l d be determined and used as a b a s i s f o r d i s c r i m i n a t i o n between the two groups. These data c o n s i s t e d of values f o r s e v e r a l features d e f i n e d and measured on a 64 p i x e l segment of the image l i n e surrounding each detected s i g n a l peak. These segments were i s o l a t e d and analyzed by the d i g i t a l s i g n a l processor of the C e l l Analyzer. Only one segment was analyzed per " h i t " recorded by the c e l l d e t e c t i o n procedure. I f an o b j e c t was detected on up to 3 adjacent image l i n e s , the l i n e c o n t a i n i n g the h i g h e s t s i g n a l peak was used f o r the a n a l y s i s . C h a r a c t e r i z a t i o n of s i g n a l s from c e l l s and n o n - c e l l s r e q u i r e d each detected o b j e c t to be c l a s s i f i e d . These c l a s s i f i c a t i o n s were de f i n e d on the b a s i s of manual observations made on two separate occasions. The f i r s t manual r e v i s i t was performed immediately a f t e r the f l a s k had been scanned, w h i l e the second r e v i s i t occurred a f t e r the c e l l s had been allowed a 2-3 97 day growth p e r i o d a t 37 °C. The f i n a l c l a s s i f i c a t i o n of each obje c t was deduced from these two sets of observations according to the scheme o u t l i n e d i n Table I. Objects were assigned to one of 4 groups ( c e l l , doublet, d e b r i s , and unsure) i n the f i n a l c l a s s i f i c a t i o n . Objects w i t h the "unsure" c l a s s i f i c a t i o n c o n s t i t u t e d only a very small p a r t (3% or l e s s ) of any sample, and were not used i n f u r t h e r analyses of the data. Doublets were a l s o not used f o r f u r t h e r a n a l y s i s , since t h e i r s i g n a l s c o u l d be very s i m i l a r to or very d i f f e r e n t from those of s i n g l e c e l l s , depending on t h e i r o r i e n t a t i o n r e l a t i v e to the l i n e a r sensor. Indeed, most doublets were a u t o m a t i c a l l y given a s p e c i a l l a b e l by the C e l l Analyzer on the b a s i s of the m u l t i p l e s i g n a l peaks they produced, e i t h e r on the same or on adjacent image l i n e s . This l a b e l was assigned regardless of the c l a s s i f i c a t i o n made by the r e c o g n i t i o n algorithm. 5 . 3 C e l l Signal Features The values of 18 s i g n a l features were c a l c u l a t e d f o r each " h i t " recorded by the c e l l d e t e c t i o n procedure. These features were d e r i v e d from a l a r g e r set o r i g i n a l l y developed f o r the r e c o g n i t i o n of CHO c e l l s ( J a g g i et a l , 1986; Poon et a l , 1987), which produce s i g n a l s s i m i l a r to those of V79 c e l l s . The s e l e c t i o n of the 18 feat u r e s had been based on t h e i r a b i l i t y to d i s c r i m i n a t e between c e l l s and d e b r i s , and a l s o to some degree on the computational time r e q u i r e d to c a l c u l a t e them. The o r i g i n a l f e a t u r e s et had in c l u d e d measurements i n both the space and the frequency domains. Information i n the frequency domain had been i n v e s t i g a t e d because of demonstrated d i s c r i m i n a t i n g c a p a b i l i t i e s i n some ob j e c t r e c o g n i t i o n a p p l i c a t i o n s . These in c l u d e r e c o g n i t i o n of alphanumeric characters and other b i n a r y images (Casasent, 1985) and, more s i g n i f i c a n t l y , T a b l e I . Manual c l a s s i f i c a t i o n scheme f o r V79 c e l l r e c o g n i t i o n a l g o r i t h m development day 0 c l a s s day 1 o r 2 c l a s s f i n a l c l a s s i f i c a t i o n ^ any any d e b r i s d o u b l e t any c e l l , d o u b l e t , o r u n s u r e c e l l d e b r i s no o b j e c t c e l l o r c o l o n y u n s u r e no o b j e c t c e l l d e b r i s d e b r i s d o u b l e t u n s u r e u n s u r e i . e . t h e f i n a l manual c l a s s i f i c a t i o n a s s i g n e d t o an o b j e c t was b a s e d on a c o m b i n a t i o n o f t h e o b s e r v a t i o n s made d u r i n g two manual r e v i s i t s (on day 0 and on day 1 o r 2 ) o f the d e t e c t e d o b j e c t s 99 d i s c r i m i n a t i o n of normal and abnormal c e l l s i n c e r v i c a l c y t o l o g i c a l samples (Per n i c k et a l , 1978; H u t z l e r , 1977), counting of red blood c e l l s (Carlson and Lee, 1978), chromosome a n a l y s i s ( H u t z l e r , 1977), and c h a r a c t e r i z a t i o n of d i f f e r e n t species of diatoms (Almeida and F u j i i , 1979). The d i s c r i m i n a t i n g power of features i n the frequency domain of l i n e scans taken by the C e l l Analyzer d i d not, however, prove to be outstanding f o r d i s t i n g u i s h i n g c e l l s from d e b r i s . In a d d i t i o n , they r e q u i r e d a great deal of computation time, s i n c e the F o u r i e r s p e c t r a had to be generated a n a l y t i c a l l y ( i . e . image data had to be gathered i n the space domain so that the l o c a t i o n of each c e l l c o u l d be p r e c i s e l y determined) . The frequency domain features were ther e f o r e among those t h a t were not i n c l u d e d i n the f i n a l f e a t u r e set. C a l c u l a t i o n of the 18 s i g n a l features s e l e c t e d by the p r e l i m i n a r y f e a t u r e e v a l u a t i o n was performed i n r e a l time as each f l a s k was scanned. Table I I gives a d e s c r i p t i o n of the features ( i n d i v i d u a l features w i l l g e n e r a l l y be r e f e r r e d to by t h e i r f e a t u r e numbers or t h e i r abbreviated d e s c r i p t i o n s , as l i s t e d i n the t a b l e ) . Some of the f e a t u r e s are f u r t h e r i l l u s t r a t e d i n Figure 32, which shows how they were measured on the s i g n a l of a t y p i c a l V79 c e l l . C e r t a i n features measured the geometry of the c e l l s i g n a l , w h i le others described more general p r o p e r t i e s r e l a t i n g to the o v e r a l l s i g n a l form (e.g. feature f i 4 , which measured the c o r r e l a t i o n between the detected s i g n a l and a predefined "master" c e l l s i g n a l ) . The "master" c e l l s i g n a l used to c a l c u l a t e features f i 3 , f i 4 , and f 1 5 was generated by manually s e l e c t i n g approximately 30 c e l l s and averaging t h e i r s i g n a l s r e l a t i v e to background. This was done p r i o r to c o l l e c t i n g data f o r generating the r e c o g n i t i o n algorithm. Some of the s i g n a l features described i n Table I I had considerable power to d i s c r i m i n a t e between V79 c e l l s and the types of n o n - c e l l s i g n a l s Table II. C e l l S i g n a l Features Feature Number Feature D e s c r i p t i o n Abbreviated D e s c r i p t i o n U n i t s fx d i s t a n c e between adjacent minima f 2 maximum peak height between minima ( r e l a t i v e to 0 i n t e n s i t y ) f 3 maximum peak height between minima ( r e l a t i v e to background) f 4 average depth of minima r e l a t i v e to background f s peak height r e l a t i v e to average minima depth f 6 b r i g h t n e s s around c e l l r e l a t i v e to background f 7 peak width at h a l f height r e l a t i v e to background f 8 sum of minima widths at h a l f the average minima depth r e l a t i v e to background f . f s / f s f i o ^3/^5^8 f l l ^3^?/^4^8 f 1 2 sum of absolute gradients along 64 p i x e l segment centered on peak f 1 3 c o r r e l a t i o n of sample s i g n a l w i t h "master" c e l l s i g n a l f 1 4 c o r r e l a t i o n of (sample-background) w i t h "master" c e l l s i g n a l f l 5 f l 4 / f 3 f 1 6 number of p i x e l s above d e t e c t i o n t h r e s h o l d i n the peak f 1 7 number of p i x e l s between background l e v e l s b o rdering adjacent minima peak width absolute peak height peak h e i g h t depth peak to trough b r i g h t r i m peak width a t h a l f h e i g ht minima width p i x e l s grey l e v e l s grey l e v e l s grey l e v e l s grey l e v e l s grey l e v e l s p i x e l s p i x e l s sum of gradients absolute c o r r e l a t i o n c o r r e l a t i o n peak width above t h r e s h o l d c e l l width f 1 8 number of p i x e l s between peak center and background outside one of the minima c e l l (always measured i n same d i r e c t i o n ) h a l f width grey l e v e l p i x e l grey l e v e l 2 grey l e v e l 2 grey l e v e l s p i x e l s p i x e l s p i x e l s 101 I*fl8+I Figure 32. algorithm, features. Geometrical signal features used i n the c e l l recognition Refer to Table II for further d e f i n i t i o n of these and other 102 t y p i c a l l y encountered i n a t i s s u e c u l t u r e f l a s k . Among these fe a t u r e s were f 3 and f 1 4 . P l o t s of t h e i r d i s t r i b u t i o n s f o r c e l l s and d e b r i s are shown i n Figures 33a,b. In c o n t r a s t , features such as f t and f 4 (Figures 33c,d) d i d not have a l o t of d i s c r i m i n a t i n g power on t h e i r own, but were s t i l l u s e f u l when combined w i t h other features i n a r e c o g n i t i o n a l g o r i t h m . Most of the 18 fe a t u r e s c a l c u l a t e d by the C e l l Analyzer had d i s c r i m i n a t i n g powers i n between the two extremes shown i n Figure 33. 5 .4 Feature S e l e c t i o n and Discr i m i n a n t Function A n a l y s i s L i n e a r d i s c r i m i n a n t f u n c t i o n a n a l y s i s was used to generate a c l a s s i f i e r to d i s t i n g u i s h between c e l l and n o n - c e l l s i g n a l s on the b a s i s of fea t u r e values c a l c u l a t e d by the C e l l Analyzer. This type of a n a l y s i s i s i n common use f o r object c l a s s i f i c a t i o n , and had p r e v i o u s l y been employed i n s t u d i e s of CHO c e l l r e c o g n i t i o n (Poon et a l , 1987). L i n e a r d i s c r i m i n a n t f u n c t i o n a n a l y s i s i s based on Bayes' c l a s s i f i c a t i o n r u l e , which s t a t e s that an o b j e c t should be assigned to the group which has the hig h e s t c o n d i t i o n a l p r o b a b i l i t y of c o n t a i n i n g i t f o r a given set of featu r e v a l u e s . The c l a s s i f i e r obtained by app l y i n g Bayes' r u l e to a given data s et y i e l d s the minimum p o s s i b l e t o t a l c l a s s i f i c a t i o n e r r o r f o r that data s e t . I f c e r t a i n assumptions are made about the nature of the feature d i s t r i b u t i o n s w i t h i n a data s e t , the c o n d i t i o n a l p r o b a b i l i t y f o r each group can be c a l c u l a t e d r e l a t i v e l y simply u s i n g a l i n e a r equation. The assumptions that must be made are: (1) each feature i s normally d i s t r i b u t e d f o r each group and (2) the covariance matrices are the same f o r a l l the groups. Even i f these c r i t e r i a are not r i g o r o u s l y met, the l i n e a r form of the c l a s s i f i c a t i o n r u l e u s u a l l y performs q u i t e w e l l (James, 1985). Moreover, i t i s r e l a t i v e l y easy to compute. 1 0 3 0.14 £ 0.12 I 0.10 2. 0.08 © 0.06-1 | 0.04 1 0.02 0.00 L (a) • non-cells • cells © 0.10 0.08-| 0.06-1 in (b) a non-cells • cells i l^" i fT^ i 11111 0 100 200 300 400 500 600 700 40 80 120 f3 - peak height (grey levels) f 1 4 - correlation (grey levels) 0.10 • non-cells • cells 10 20 30 40 50 ([ - peak width (pixels) f f f r 0 10 20 30 40 50 60 70 f 4 - depth (grey levels) Figure 33. D i s t r i b u t i o n s o f c e l l and n o n - c e l l f e a t u r e v a l u e s f o r : (a) f 3 (peak h e i g h t r e l a t i v e t o b a c k g r o u n d ) , (b) f 1 4 ( c o r r e l a t i o n o f d e t e c t e d s i g n a l r e l a t i v e t o bac k g r o u n d w i t h a p r e - d e f i n e d "master" c e l l s i g n a l ) , ( c ) f ! (peak w i d t h ) , (d) f 4 (average d e p t h o f minima r e l a t i v e t o ba c k g r o u n d ) . 104 The l i n e a r c l a s s i f i c a t i o n r u l e can be expressed i n the form: assign object to group i if c.n+c..f,+c.-f-+. . .+c. f > c.r+c..f+c.0f.+ . . .+ci f 10 i l 1 12 2 in n jO j l 1 j2 2 jn n for all j ? I (5.1) where n i s the t o t a l number of features and the c o e f f i c i e n t s c ^ are constants t h a t m u l t i p l y the feat u r e v a r i a b l e s f"L. A separate s et of c o e f f i c i e n t s i s c a l c u l a t e d f o r each group by a n a l y z i n g the feature d i s t r i b u t i o n s of a " l e a r n i n g " data set ( s p e c i f i c formulae f o r the e v a l u a t i o n of the d i s c r i m i n a n t f u n c t i o n parameters are given i n the Appendix, S e c t i o n 10.1). Once computed, the c o e f f i c i e n t s can be used to a s s i g n u n c l a s s i f i e d objects to groups on the b a s i s of t h e i r f e a t u r e values. I f there are only two groups to which objects are to be assigned, the l i n e a r c l a s s i f i c a t i o n r u l e can be represented i n a simpler form: assign object to group 1 if C,f ,+C0f,+ . . .+C f > Cn 1 1 2 2 n n D otherwise assign object to group 2 (5.2) where C = c-c.v and C_, = c -c,„ = -Cn. The sum i s r e f e r r e d to as the K IK ZK D 20 10 0 d i s c r i m i n a n t score, and the constant as the d e c i s i o n boundary. This form of the c l a s s i f i c a t i o n r u l e can be i n t e r p r e t e d as a hyperplane s e p a r a t i n g the two groups i n the feature space. The d i s c r i m i n a n t score f o r an ob j e c t i s the p r o j e c t i o n of i t s feature v e c t o r onto a l i n e p e r p e n d i c u l a r to the hyperplane. Adjustment of any of the c o e f f i c i e n t s C , C2, . . . , changes the o r i e n t a t i o n of the hyperplane, while adjustment of CQ a l t e r s i t s l o c a t i o n i n the feature space. Changing the value of C i s a simple 105 means of a d j u s t i n g the r e l a t i v e magnitudes of the f a l s e p o s i t i v e and f a l s e negative e r r o r s . This can prove u s e f u l when e i t h e r of these e r r o r s has to be c a r e f u l l y r e g u l a t e d , as i s the case f o r c e l l r e c o g n i t i o n . Changing C D from the optimum value does, however, increase the t o t a l c l a s s i f i c a t i o n e r r o r . Such adjustments must ther e f o r e be made w i t h d i s c r e t i o n . The computations necessary f o r implementation of the two-group form of the l i n e a r d i s c r i m i n a n t f u n c t i o n , i n c l u d i n g c a l c u l a t i o n of the feature v a l u e s , c o u l d be c a r r i e d out i n r e a l time by the d i g i t a l s i g n a l processor of the C e l l Analyzer. Objects could t h e r e f o r e be c l a s s i f i e d i n t o e i t h e r " c e l l " (group 1) or " n o n - c e l l " (group 2) c a t e g o r i e s as they were detected. The BMDP s t a t i s t i c a l software package (BMDP S t a t i s t i c a l Software, Los Angeles, C a l i f o r n i a ) was used to generate d i s c r i m i n a n t f u n c t i o n c o e f f i c i e n t s from l e a r n i n g s e t s . This software package can a l s o perform a stepwise a n a l y s i s of the data p r i o r to generating the d i s c r i m i n a n t f u n c t i o n ( J e n n r i c h and Sampson, 1983). The purpose of such a procedure i s to s e l e c t those fe a t u r e s that provide the best d i s c r i m i n a t i o n , while d i s r e g a r d i n g those that make l i t t l e or no c o n t r i b u t i o n to the c l a s s i f i c a t i o n process. Features are s e l e c t e d on the b a s i s of t h e i r a b i l i t y to separate the groups i n f e a t u r e space. The methods used i n t h i s s e l e c t i o n procedure are d e s c r i b e d i n greater d e t a i l i n the Appendix, S e c t i o n 10.1. To generate the r e c o g n i t i o n a l g o r i t h m f o r V79 c e l l s , a stepwise d i s c r i m i n a n t a n a l y s i s was performed on s e v e r a l independent data s e t s , each c o n s i s t i n g of 1000 to 2000 o b j e c t s . The data c o n s i s t e d of f e a t u r e values c a l c u l a t e d f o r a l l detected o b j e c t s , except f o r those that had been manually c l a s s i f i e d as "unsure" or "doublet". The f e a t u r e s s e l e c t e d by the stepwise d i s c r i m i n a n t a n a l y s i s v a r i e d somewhat amongst the d i f f e r e n t data s e t s , but c e r t a i n important features were i d e n t i f i a b l e . Among these were fi4> f i 6 . a n c * f i 7 - Sometimes f 1 3 was s e l e c t e d i n place of f i 4 , but both 106 were never i n c l u d e d simultaneously i n the same set of s e l e c t e d f e a t u r e s . This was because h i g h l y c o r r e l a t e d f e a t u r e s , as these two were, are not accepted together by the stepwise a n a l y s i s . Features f 2 and f 3 were a l s o never s e l e c t e d together f o r the same reason. Another f e a t u r e that e x h i b i t e d a h i g h degree of c o r r e l a t i o n w i t h other f e a t u r e s was f 4 , but i n t h i s case the c o r r e l a t i o n was w i t h a l i n e a r combination of s e v e r a l other f e a t u r e s . As a r e s u l t , f 4 was r a r e l y s e l e c t e d by the stepwise a n a l y s i s . A f t e r the stepwise a n a l y s i s had been performed on s e v e r a l data s e t s , the f i n a l f e a t u r e s to be used f o r the generation of the l i n e a r d i s c r i m i n a n t f u n c t i o n were chosen. These features were s e l e c t e d p r i m a r i l y on the b a s i s of the r e s u l t s of the stepwise a n a l y s i s , but the r e p r o d u c i b i l i t y of the featu r e d i s t r i b u t i o n s from f l a s k to f l a s k was a l s o considered. The set of features f i n a l l y s e l e c t e d i n c l u d e d a l l except f 6 (which d i d not produce c o n s i s t e n t d i s t r i b u t i o n s from f l a s k to f l a s k ) and f 2 , f 4 , and f i 3 , (which c o r r e l a t e d h i g h l y w i t h other f e a t u r e s ) . Features f 3 and f 1 4 were chosen i n s t e a d of f 2 and f 1 3 because they are measured r e l a t i v e to background i n t e n s i t y r a t h e r than as absolute i n t e n s i t i e s . Feature values measured us i n g absolute i n t e n s i t i e s were l e s s c o n s i s t e n t because of s l i g h t f l u c t u a t i o n s i n background l i g h t l e v e l s between f l a s k s . Using the s e l e c t e d f e a t u r e s , the l i n e a r d i s c r i m i n a n t f u n c t i o n f o r the V79 c e l l r e c o g n i t i o n a l g o r i t h m was c a l c u l a t e d from a l e a r n i n g set c o n s i s t i n g of 2200 objects (62% c e l l s and 38% d e b r i s ) . The c a l c u l a t e d d i s c r i m i n a n t weights, CK, were s c a l e d and rounded o f f to one or two s i g n i f i c a n t f i g u r e s f o r implementation by the d i g i t a l s i g n a l processor of the C e l l Analyzer, which uses i n t e g e r a r i t h m e t i c . In order to achieve the minimum t o t a l c l a s s i f i c a t i o n e r r o r a f t e r t h i s rounding o f f procedure, i t was necessary to r e - a d j u s t the l o c a t i o n of the d e c i s i o n boundary, C^. A f t e r t h i s had been done, the t o t a l c l a s s i f i c a t i o n e r r o r of the 107 d i s c r i m i n a n t f u n c t i o n was 9.8% f o r the l e a r n i n g set (as compared to 9.3% before any c o e f f i c i e n t s had been rounded o f f ) . The s e p a r a t i o n of c e l l s and n o n - c e l l s by the d i s c r i m i n a n t f u n c t i o n i s i l l u s t r a t e d i n the histogram of Figure 34, which shows the d i s t r i b u t i o n of d i s c r i m i n a n t scores f o r these two groups i n the l e a r n i n g s e t . The l o c a t i o n of the d e c i s i o n boundary f o r minimum t o t a l e r r o r i s a l s o shown. False p o s i t i v e and f a l s e negative e r r o r s f o r these data (using the rounded o f f d i s c r i m i n a n t weights) were 9.4% and 10.6%, r e s p e c t i v e l y . The f a l s e p o s i t i v e e r r o r was t h e r e f o r e higher than the s p e c i f i e d 5% l i m i t . To lower the f a l s e p o s i t i v e e r r o r , the l o c a t i o n of the d e c i s i o n boundary, C^, had to be adjusted away from i t s optimum value f o r minimum t o t a l e r r o r . The e f f e c t of changing C D on the e r r o r r a t e s f o r the l e a r n i n g set i s shown i n Figure 35, as i s the corresponding e f f e c t on the p r o p o r t i o n of c e l l s and the p r o p o r t i o n of d e b r i s c o r r e c t l y c l a s s i f i e d . As i n d i c a t e d by the f i g u r e , a f a l s e p o s i t i v e e r r o r of c l o s e to 5% could be achieved at a value of CD - 100 without d r a s t i c a l l y i n c r e a s i n g the t o t a l c l a s s i f i c a t i o n e r r o r . A f t e r adjustment of the d e c i s i o n boundary, the f a l s e p o s i t i v e e r r o r f o r the l e a r n i n g set was 5.3%, which was acceptably c l o s e to the s p e c i f i e d 5% l i m i t . The accompanying t o t a l c l a s s i f i c a t i o n e r r o r was 12.1%, approximately 2 percentage p o i n t s higher than the optimum value of 9.8%. This increase i n the t o t a l e r r o r was r e f l e c t e d i n a f a i r l y l a r g e increase i n the f a l s e negative e r r o r , which almost doubled to 20.9%. However, because the data set contained more c e l l s than d e b r i s , the increased f a l s e negative e r r o r s t i l l y i e l d e d a c o r r e c t c l a s s i f i c a t i o n of 85% of the c e l l s . The performance of the d i s c r i m i n a n t f u n c t i o n was evaluated f o r s e v e r a l independent t e s t data sets (see Table I I I ) . In terms of f a l s e p o s i t i v e e r r o r , the performance of the d i s c r i m i n a n t f u n c t i o n f o r the t e s t sets was as good as or b e t t e r than f o r the l e a r n i n g s e t . The p r o p o r t i o n of 108 CD o O o a -1000 -500 0 500 discriminant score 1000 F i g u r e 34. D i s t r i b u t i o n o f d i s c r i m i n a n t s c o r e s i n the l e a r n i n g s e t u s e d t o g e n e r a t e t he l i n e a r d i s c r i m i n a n t f u n c t i o n f o r c e l l r e c o g n i t i o n . The i l l u s t r a t e d d e c i s i o n boundary g i v e s the minimum t o t a l e r r o r f o r c l a s s i f i c a t i o n o f c e l l s and n o n - c e l l s by the d i s c r i m i n a n t f u n c t i o n . 109 decision boundary 1 0 0 80-(b) % false positive error % false negative error % total error -600 -400 -200 0 200 400 decision boundary 600 Figure 35. E f f e c t of a l t e r i n g the p o s i t i o n of the de c i s i o n boundary on discriminant function performance for the learning set. (a) e f f e c t on the percentage of c e l l s and non-cells c o r r e c t l y c l a s s i f i e d . (b) e f f e c t on fa l s e p o s i t i v e , f a l s e negative, and t o t a l error rates. Table I I I . D i s c r i m i n a n t f u n c t i o n performance f o r V79 c e l l r e c o g n i t i o n Data Set no. of obj ects % d e b r i s % c e l l s c o r r e c t % d e b r i s c o r r e c t % f a l s e p o s i t i v e % t o t a l e r r o r l e a r n i n g 2191 38.6 85.4 92.1 5.3 12.1 t e s t #1 3248 6.7 79.0 91.3 0.8 20.2 t e s t #2 2369 25.5 79.3 87.2 5.2 18.7 t e s t #3 1599 34.0 68.9 93.7 4.5 22.7 t e s t #4 928 28.7 70.5 92.1 4.3 23.3 (j, ± a f o r a l l sets combined 2067 ± 869 26.7 ± 12.3 76.6 ± 6.8 91.3 ± 2.4 4.0 ± 1.9 19.4 ± 4.5 I l l m i s c l a s s i f i e d c e l l s and the t o t a l error were, however, somewhat higher for most of the test sets. In general, higher error rates would be expected fo r the independent t e s t sets, since they were not used d i r e c t l y f or the generation of the discriminant function. On the basis of the t e s t sets, a t o t a l c l a s s i f i c a t i o n error of c l o s e r to 20% (as compared to the 12% t o t a l error obtained for the learning set) would be expected i n the a p p l i c a t i o n of the discriminant function to the objects detected i n any given f l a s k . The increased error appears to a r i s e c o n s i s t e n t l y from m i s c l a s s i f i c a t i o n of c e l l s rather than m i s c l a s s i f i c a t i o n of debris, suggesting that the c h a r a c t e r i s t i c c e l l s i g n a l i s subject to some degree of day-to-day v a r i a b i l i t y despite e f f o r t s to maintain consistency i n c e l l preparation procedures and adjustment of focus during scanning. Nevertheless, since i t was the f a l s e negative rather than the f a l s e p o s i t i v e e r r o r that was affected, these v a r i a t i o n s could be tolerated. 5.5 P r e - c l a s s i f i c a t i o n of Objects 5.5.1 Imposition of Bounds on Feature Values A f t e r the c o e f f i c i e n t s of the l i n e a r discriminant function had been established, a means of p r e - c l a s s i f y i n g objects p r i o r to a p p l i c a t i o n of the discriminant function was investigated (Spadinger et a l , 1990). This pre-c l a s s i f i c a t i o n was performed by imposing upper and lower bounds on feature values. Those objects with feature values f a l l i n g within the range s p e c i f i e d by the bounds would be accepted for further c l a s s i f i c a t i o n by the discriminant function, while a l l other objects would immediately be c l a s s i f i e d as non-cells. In t h i s way, features such as f 4 could be given some di s c r i m i n a t i n g power. The use of upper and lower bounds ei t h e r decreases, or has no e f f e c t on, the f a l s e p o s i t i v e error r e l a t i v e to the 112 performance of the d i s c r i m i n a n t f u n c t i o n alone. Furthermore, the t o t a l c l a s s i f i c a t i o n e r r o r does not n e c e s s a r i l y increase when improvements i n the f a l s e p o s i t i v e e r r o r are seen. Bounds were i n i t i a l l y s e l e c t e d f o r each fea t u r e by examining i t s d i s t r i b u t i o n f o r c e l l s and de b r i s and determining the percentage of each c l a s s of ob j e c t s f a l l i n g w i t h i n a s p e c i f i e d range of values. Bounds could be set so th a t the m a j o r i t y of c e l l s f e l l w i t h i n t h e i r l i m i t s , w h i l e as much d e b r i s as p o s s i b l e f e l l o u t s i d e , as i l l u s t r a t e d f o r features f 1 1 ( f 1 2 and f 1 7 i n Figure 36. Various combinations of feat u r e bounds were a p p l i e d to s e v e r a l independent data s e t s . The number of c e l l s and n o n - c e l l s e l i m i n a t e d by the bounds f o r each f e a t u r e , as w e l l as the t o t a l number e l i m i n a t e d by the combined a c t i o n of a l l the f e a t u r e bounds, was determined. From these r e s u l t s , i t was found that bounds imposed on d i f f e r e n t features g e n e r a l l y d i d not e l i m i n a t e the same o b j e c t s . The t o t a l number of ob j e c t s f a l l i n g outside feature bounds f o r a given data set was th e r e f o r e l a r g e r (by a f a c t o r of 2 or 3) than the number f a l l i n g outside the bounds f o r any i n d i v i d u a l f e a t u r e . A f t e r t e s t i n g the e f f e c t s of both r e l a t i v e l y s t r i n g e n t bounds, where as many as 5% of c e l l s and 40% of de b r i s f e l l o utside the bounds f o r some fe a t u r e s , and more r e l a x e d bounds, where l e s s than 1% of c e l l s and 15% of de b r i s t y p i c a l l y f e l l outside the feat u r e bounds, i t was found that the more r e l a x e d bounds provided a b e t t e r o v e r a l l improvement i n the performance of the r e c o g n i t i o n algorithm. Bounds th a t were too s t r i n g e n t r e s u l t e d i n the m i s c l a s s i f i c a t i o n of too many c e l l s without producing markedly b e t t e r f a l s e p o s i t i v e r a t e s than were achieved w i t h more r e l a x e d bounds. In comparison to the performance of the d i s c r i m i n a n t f u n c t i o n alone, a p p l i c a t i o n of s t r i n g e n t f e a t u r e bounds r e s u l t e d i n a 0.5-2.5 percentage p o i n t decrease i n the f a l s e p o s i t i v e e r r o r and a 4-6 percentage 1 1 3 a o cd 0.25 0.20 I 0.15 (a) a non-cells • cells upper bound (lower bound at 0) 0 50 PTT| I I I r p i r i n 0.10 0.08-t ! 0.06-1 fl o • rt o u 100 150 200 250 300 350 0.04-0.02-0.00 (b) i lower bound • non-cells • cells upper bound rT-r 100 200 300 400 500 600 700 fig-sum of gradients (grey level/pixel) • I 0.14 § 0.02 ^ 0.00 10 20 30 40 50 60 f 1 7 - cell width (pixels) 70 F i g u r e 36. Use of upper and lower bounds on feature values to i d e n t i f y some non-cell objects p r i o r to a p p l i c a t i o n of the l i n e a r discriminant function. Objects outside the upper and lower bounds are c l a s s i f i e d as non-cells, while objects inside the bounds are analyzed further with the discriminant function. 114 p o i n t increase i n the number of c e l l s m i s c l a s s i f i e d as d e b r i s . The more r e l a x e d f e a t u r e bounds, on the other hand, increased the m i s c l a s s i f i c a t i o n r a t e f o r c e l l s by l e s s than 1 percentage p o i n t w h i l e improving the f a l s e p o s i t i v e e r r o r to a degree s i m i l a r to that obtained u s i n g s t r i n g e n t bounds. On c l o s e examination of the a c t i o n of the feat u r e bounds i n r e l a t i o n to the a c t i o n of the l i n e a r d i s c r i m i n a n t f u n c t i o n , i t was found that a l a r g e p r o p o r t i o n (50-100%, depending on the feat u r e and the data set) of the d e b r i s e l i m i n a t e d by the feat u r e bounds was c o r r e c t l y c l a s s i f i e d by the d i s c r i m i n a n t f u n c t i o n . However, bounds f o r c e r t a i n f e a t u r e s c o n s i s t e n t l y e l i m i n a t e d mostly n o n - c e l l s that were m i s c l a s s i f i e d by the d i s c r i m i n a n t f u n c t i o n . The most prominent among these features was f n , f o r which, on average, 65% of the debris objects e l i m i n a t e d by the feat u r e bounds would not have been c o r r e c t l y c l a s s i f i e d by the d i s c r i m i n a n t f u n c t i o n . The success of the feat u r e bounds i n t h i s case was due to the f a c t that the n o n - c e l l f e a t u r e d i s t r i b u t i o n f o r f l x i s very skewed. Bounds on some other features (most notably f 4 , f i o . a n d f i 2 ) e l i m i n a t e d j u s t as many d e b r i s o b j e c t s m i s c l a s s i f i e d by the d i s c r i m i n a n t f u n c t i o n as f x x d i d , but these bounds a l s o e l i m i n a t e d a r e l a t i v e l y l a rge number of d e b r i s o b j e c t s that were c o r r e c t l y c l a s s i f i e d by the d i s c r i m i n a n t f u n c t i o n . The e f f e c t of p r e - c l a s s i f y i n g objects through the use of feat u r e bounds i s shown i n Table IV, which compares the a c t i o n of the l i n e a r d i s c r i m i n a n t f u n c t i o n alone to the combined a c t i o n of f e a t u r e bounds and the d i s c r i m i n a n t f u n c t i o n . P r e - c l a s s i f i c a t i o n of objects r e s u l t e d i n a d e f i n i t e improvement i n the f a l s e p o s i t i v e e r r o r r a t e without s e r i o u s l y c o n t r i b u t i n g to c e l l m i s c l a s s i f i c a t i o n . Furthermore, the t o t a l e r r o r e i t h e r decreased or was not a f f e c t e d by the feat u r e bounds. Thus, upper and lower bounds on feat u r e values proved to be a b e n e f i c i a l a d d i t i o n to the c e l l r e c o g n i t i o n procedure, and were permanently i n c o r p o r a t e d i n t o i t . 115 Table I V . E f f e c t of feature bounds on r e c o g n i t i o n a l g o r i t h m performance % c e l l s % d e b r i s % f a l s e % t o t a l c o r r e c t c o r r e c t p o s i t i v e e r r o r Data Set l no 1 , bounds bounds 1 no | bounds| 1 bounds 1 no | bounds| j bounds 1 no 1 . bounds bounds l e a r n i n g 85.4 1 J 84.8 1 92.1 j 94.2 » ! 4.0 12.1 t e s t #1 79.0 j 78.8 91.3 j 95.9 0.8 j 0.4 20.2 j 20.0 t e s t #2 79.3 j 78.8 87.2 j 88.7 » ! 4.7 18.7 j 18.7 t e s t #3 68.9 j 68.5 93.7 j 97.4 1.9 22.7 j 21.7 t e s t #4 70.5 j 69.9 I 92.1 J 1 95.5 » ! 2.5 23.3 J 22.7 1 fj. ± a f o r 76.6 T | 76.2 T 91.3 | 94.3 2.7 19.4 T | 19.0 a l l sets ± ± + + combined 6.8 j 6.8 1 2 , j 1 3.4 1 1.7 4.5 j 4 , l 116 5.5.2 E x c l u s i o n of Near Neighbours Some " p r e - c l a s s i f i c a t i o n " of detected objects c o u l d a l s o be achieved by the e x c l u s i o n of near neighbours from data sets c o l l e c t e d by the c e l l d e t e c t i o n procedure. This " p r e - c l a s s i f i c a t i o n " occurred p r i m a r i l y through the e l i m i n a t i o n of la r g e pieces of d e b r i s , which co u l d be detected on s e v e r a l , not n e c e s s a r i l y adjacent, l i n e scans. These o b j e c t s would be e l i m i n a t e d from the data set due to the p r o x i m i t y of the " h i t s " that they generated. Table V summarizes the e f f e c t of near neighbour e x c l u s i o n on c e l l r e c o g n i t i o n (where the r e c o g n i t i o n a l g o r i t h m c o n s i s t s of the l i n e a r d i s c r i m i n a n t f u n c t i o n plus the feature bounds). A 25 /xm e x c l u s i o n distance was used. This distance was small enough to r e s u l t i n the e l i m i n a t i o n of only 3-5% of the c e l l s from the data s e t s , while s t i l l e l i m i n a t i n g 40-50% of the d e b r i s . Larger e x c l u s i o n distances ( i . e . up to 600 /xm) a f f e c t e d r e c o g n i t i o n e r r o r r a t e s and the p r o p o r t i o n of de b r i s i n the data sets i n v i r t u a l l y the same way as the 25 /xm d i s t a n c e , even though they a l s o reduced the t o t a l s i z e of the data set considerably. A f t e r a p p l y i n g the e x c l u s i o n procedure, the f a l s e p o s i t i v e e r r o r was reduced by 0.5-1 percentage p o i n t f o r most of the data s e t s . Other measures of performance (e.g. t o t a l e r r o r and percentage of c e l l s and debr i s c o r r e c t l y c l a s s i f i e d ) were not a f f e c t e d as c o n s i s t e n t l y . This was to be expected, however, since these q u a n t i t i e s do not depend on the composition ( i . e . number of c e l l s i n r e l a t i o n to d e b r i s ) of the data s e t . The f a l s e p o s i t i v e e r r o r s generated a f t e r e x c l u s i o n of near neighbours compared very w e l l to those of a human observer. Thus, near neighbour e x c l u s i o n appeared to be a u s e f u l means of reducing r e c o g n i t i o n e r r o r s , even when i t was not necessary f o r o b t a i n i n g a data s et of w e l l -separated c e l l s . In the rare instances that the l a t t e r s i t u a t i o n a r i s e s , Table V. E f f e c t of p r o x i m i t y e x c l u s i o n on r e c o g n i t i o n a l g o r i t h m performance (25 /xm e x c l u s i o n d i s t a n c e ) Data Set no. of obj ects % d e b r i s % c e l l s c o r r e c t % d e b r i s c o r r e c t % f a l s e p o s i t i v e % t o t a l e r r o r l e a r n i n g before 2191 38.6 84 8 94 2 4 0 11.7 after 1774 25.0 86 5 90 8 3 4 12.5 t e s t #1 before 3248 6.7 78 8 95 9 0 4 20.0 after 3049 4.4 79 6 93 3 0 4 19.8 t e s t #2 before 2369 25.5 78 8 88 7 4 7 18.7 after 1919 15.3 82 2 85 0 3 2 17.4 t e s t #3 before 1599 34.0 68 5 97 4 1 9 21.7 after 1236 20.0 69 5 94 7 1 9 25.5 t e s t #4 before 928 28.7 69 9 95 5 2 5 22.7 after 750 19.7 72 2 93 9 2 0 23.5 n ± a before 2067 26.7 76.2 94.3 2.7 19.0 ± 869 ±12.3 + 6.8 ± 3.4 + 1.7 ± 4.3 after 1746 16.9 78.0 91.5 2.2 19.1 ± 863 ± 7.8 ± 7.0 ± 3.9 ± 1.2 ± 4.0 118 small e x c l u s i o n d i s t a n c e s could be a p p l i e d without s e r i o u s l y reducing the number of c e l l s i n the data s e t . Nevertheless, the e x c l u s i o n procedure was not i n c o r p o r a t e d as a permanent p a r t of the r e c o g n i t i o n a l g o r i t h m , since e x c l u s i o n d i s t a n c e s had to be a d j u s t a b l e according to experimental needs. Furthermore, options to perform t h i s f u n c t i o n had already been in c o r p o r a t e d i n t o the c e l l d e t e c t i o n procedures of the CSCAN and CELREC programs. 5.6 E f f e c t of Defocusing on C e l l Recognition 5.6.1 Focus E f f e c t s on C e l l S i g n a l Features As already discussed i n S e c t i o n 4.4, the focus s e t t i n g c o u l d have s i g n i f i c a n t e f f e c t s on the form of the c e l l s i g n a l generated by the C e l l Analyzer. These changes could a l t e r the values of c e l l s i g n a l features used by the r e c o g n i t i o n algorithm, and thereby a f f e c t the performance of the a l g o r i t h m i t s e l f (Spadinger et a l , 1990). The e f f e c t of defocusing on c e l l s i g n a l f e a t u r e d i s t r i b u t i o n s was measured on i n d i v i d u a l l y focused c e l l s at focus l e v e l s ranging from 100 /jm below to 150 above optimum focus (as d e f i n e d by maximization of the s i g n a l peak h e i g h t ) . R e s u l t s f o r s e v e r a l of the 18 s i g n a l features are shown i n Figure 37. C e r t a i n f e a t u r e s , such as the peak height (f2» £3), the sum of gradients ( f i 2 ) , a n d the c o r r e l a t i o n of the detected s i g n a l w i t h a p r e - d e f i n e d "master" c e l l s i g n a l ( f i 3 , f i 4 ) , were h i g h l y focus dependent. The e f f e c t of defocusing on f 3 was seen p r e v i o u s l y i n Figures 21 and 22. Figures 37a,b i l l u s t r a t e the focus dependence f o r f 1 2 and f 1 4 . A l l of these f e a t u r e s showed sharp maxima at i d e a l focus, but diminished r a p i d l y i n value f o r both p o s i t i v e and negative defocusing. The focus dependence of f 6 ( i n t e n s i t y of b r i g h t rim) i s shown i n Figure 37c. This i s a feature that was a f f e c t e d very l i t t l e by negative 1 1 9 -5 25 20 i > i i j i i i i j r i i i | i i i i | i i r i -100 -50 0 50 100 150 (e) peak width -100 -50 0 50 100 150 50--100 -50 focus level (um) focus level (um) Figure 37. E f f e c t of defocusing on selected c e l l s i g n a l features. The ide a l focus, as defined by maximum signal peak height, i s at "0". Error bars are standard deviations. Abbreviated feature descriptions (see Table II) are given on the p l o t s . 120 defocusing, but underwent l a r g e changes at p o s i t i v e defocusings as a r e s u l t of the d e t e r i o r a t i o n of the c h a r a c t e r i s t i c c e l l s i g n a l . Features l i k e t h i s c o u l d prove u s e f u l i n more s o p h i s t i c a t e d autofocusing algorithms than the one c u r r e n t l y i n use ( S e c t i o n 4.4.1), since they i n d i c a t e the d i r e c t i o n i n which focus adjustments are re q u i r e d . Another f e a t u r e t h a t was a f f e c t e d p r i m a r i l y by p o s i t i v e defocusing was f 1 0 (Figure 37d). This feature e x h i b i t e d a sudden increase at small p o s i t i v e defocusings, then decreased r a p i d l y to a r e l a t i v e l y constant value beyond approximately 60 nm above i d e a l focus. Some fea t u r e s showed a stronger focus dependence at negative than at p o s i t i v e defocusings. These features were t y p i c a l l y those that were measured on the s i g n a l minima, which d i m i n i s h i n depth and become s l i g h t l y f a r t h e r apart at negative defocusings. Examples are fi (peak width) and f 4 (depth of minima), which are shown i n Figures 37e,f. Some focus dependence was a l s o evident a t p o s i t i v e defocusings f o r these f e a t u r e s , but i t was l e s s extreme than the negative focus dependence. 5.6.2 R e c o g n i t i o n A l g o r i t h m Performance Because the features used i n the c e l l r e c o g n i t i o n a l g o r i t h m depended on focus i n d i f f e r e n t ways, i t was d i f f i c u l t to p r e d i c t the e f f e c t of defocusing on r e c o g n i t i o n a l g o r i t h m performance on the b a s i s of i n d i v i d u a l f e a t u r e measurements alone. Nevertheless, the r e c o g n i t i o n a l g o r i t h m was expected to e x h i b i t some degree of focus t o l e r a n c e , s i n c e i t had been developed u s i n g data c o l l e c t e d under normal scanning c o n d i t i o n s , where focus was allowed to vary by approximately ±30 /xm from i d e a l l e v e l s . In order to v e r i f y the extent of focus e f f e c t s on r e c o g n i t i o n a l g o r i t h m performance, however, i t was necessary to measure them d i r e c t l y . 121 The performance of the r e c o g n i t i o n a l g o r i t h m at d i f f e r e n t focus l e v e l s was measured usi n g a method s i m i l a r to that d e s c r i b e d i n S e c t i o n 4.5 f o r the determination of step s i z e and focus e f f e c t s on c e l l d e t e c t i o n . C e l l s were p l a t e d at d e n s i t i e s of 4-5 cells/mm 2 f o r t h i s procedure so that data sets of 400-500 objects could be obtained by searching only the c e n t r a l scanning band (band 6) i n a f l a s k . Measurements were made i n the range extending from -80 to +40 fim r e l a t i v e to i d e a l focus. Too few c e l l s were detected outside t h i s range to al l o w a meaningful e v a l u a t i o n of the d i s c r i m i n a n t f u n c t i o n performance. The e x c l u s i o n procedure was not a p p l i e d during the course of these measurements, so as not to reduce the s i z e of the data set u n n e c e s s a r i l y . Furthermore, the c o n t r i b u t i o n of the e x c l u s i o n procedure to c e l l r e c o g n i t i o n i s confined to reducing the p r o p o r t i o n of de b r i s i n the data s e t , which i s not a s t r i c t l y focus-dependent v a r i a b l e . The focus dependence of the percentage of c e l l s and n o n - c e l l s c o r r e c t l y c l a s s i f i e d , the f a l s e p o s i t i v e e r r o r , and the t o t a l e r r o r i s p l o t t e d i n Figure 38. Results are shown f o r the performance of the l i n e a r d i s c r i m i n a n t f u n c t i o n alone, and f o r the complete r e c o g n i t i o n a l g o r i t h m c o n s i s t i n g of the combined a c t i o n of the f e a t u r e bounds and the d i s c r i m i n a n t f u n c t i o n . The percentage of de b r i s i n the data set i s a l s o p l o t t e d as a f u n c t i o n of the focus s e t t i n g . Changes i n t h i s value were due to focus e f f e c t s on both the number of c e l l s and the amount of debris detected at the d i f f e r e n t focus s e t t i n g s . Figure 38a shows the focus dependence of the percentage of c e l l s c o r r e c t l y c l a s s i f i e d , which was a f f e c t e d very l i t t l e by a p p l i c a t i o n of the fe a t u r e bounds throughout the f o c a l range i n which measurements were made. C e l l s were c o r r e c t l y c l a s s i f i e d 85-90% of the time w i t h i n -30 to +20 /xm of i d e a l focus. This was approximately the same as the range f o r optimum c e l l d e t e c t i o n . Outside t h i s range the f r a c t i o n of c e l l s c o r r e c t l y c l a s s i f i e d 122 % total error • without bounds • with bounds V -50 i i i i i 0 80-60-40-20-% noncells correct • without bounds • with bounds (c) % false positive • without bounds • with bounds 0 -100 • F V mm mm 80-60-40-20 0 * debris in data set 0 50 -100 -50 focus level (um) 0 50 50 Figure 38. E f f e c t of defocusing on r e c o g n i t i o n a l g o r i t h m performance, both w i t h and without the use of upper and lower bounds on fe a t u r e values to p r e - c l a s s i f y detected o b j e c t s . (a) percentage of c e l l s c o r r e c t l y c l a s s i f i e d , (b) percentage of n o n - c e l l s c o r r e c t l y c l a s s i f i e d , (c) percent f a l s e p o s i t i v e s , (d) percent t o t a l e r r o r . The v a r i a t i o n i n the f r a c t i o n of debr i s making up the data set at d i f f e r e n t focus l e v e l s i s shown i n (e). 123 dropped r a p i d l y . Thus, the r e c o g n i t i o n a l g o r i t h m d i d indeed e x h i b i t a focus t o l e r a n c e corresponding to that of the data used to generate i t . The percentage of n o n - c e l l s c o r r e c t l y c l a s s i f i e d (Figure 38b) d i d not show a strong focus dependence. I n t e r e s t i n g l y , i t was somewhat lower at focus l e v e l s c l o s e to i d e a l than at focus l e v e l s f a r t h e r away. This was not t o t a l l y unexpected, however, si n c e the most e a s i l y m i s c l a s s i f i e d n o n - c e l l o bjects are l i k e l y to be very much l i k e c e l l s i n s i z e , shape, and, t h e r e f o r e , focus dependence. Because the f e a t u r e bounds had been designed to s p e c i f i c a l l y remove de b r i s from the data set w h i le r e t a i n i n g as many c e l l s as p o s s i b l e , they a f f e c t e d the percentage of n o n - c e l l s c o r r e c t l y c l a s s i f i e d more than the percentage of c e l l s c o r r e c t l y c l a s s i f i e d . This e f f e c t was r e l a t i v e l y constant throughout most of the f o c a l range i n which measurements were made, improving the f r a c t i o n of n o n - c e l l s c o r r e c t l y c l a s s i f i e d by approximately 5 percentage p o i n t s . The f a l s e p o s i t i v e e r r o r (Figure 38c) was a l s o r e l a t i v e l y independent of the focus s e t t i n g i f feature bounds were used. I f f e a t u r e bounds were not used, however, the f a l s e p o s i t i v e e r r o r increased s h a r p l y below -60 um from i d e a l focus. This increase was not due to any l a r g e increases i n the f r a c t i o n of d e b r i s detected i n t h i s f o c a l range (Figure 38e), nor was i t due to l a r g e decreases i n the f r a c t i o n of d e b r i s c o r r e c t l y c l a s s i f i e d . Rather, i t was caused p r i m a r i l y by the s m all number of c e l l s c o r r e c t l y c l a s s i f i e d . Small improvements i n the f r a c t i o n of d e b r i s s i g n a l s c o r r e c t l y c l a s s i f i e d c o u l d t h e r e f o r e have la r g e e f f e c t s on the f a l s e p o s i t i v e e r r o r r a t e , e x p l a i n i n g the i n f l u e n c e of the feature bounds i n t h i s f o c a l range. The focus dependence f o r the t o t a l c l a s s i f i c a t i o n e r r o r i s shown i n Figure 38d. The m i s c l a s s i f i c a t i o n of c e l l s outside the -40 to +20 um f o c a l range a f f e c t s the t o t a l e r r o r r a t e , which was r e l a t i v e l y s t a b l e at 10-15% 124 w i t h i n t h i s range but increased r a p i d l y outside i t . Feature bounds had l i t t l e e f f e c t on the t o t a l e r r o r because i t was dominated more by the e r r o r due to m i s c l a s s i f i c a t i o n of c e l l s than by the e r r o r due to m i s c l a s s i f i c a t i o n of d e b r i s . I t t h e r e f o r e appeared that the r e c o g n i t i o n a l g o r i t h m c o u l d be r e l i e d upon to perform w e l l i n terms of f a l s e p o s i t i v e e r r o r r a t e s a t focus l e v e l s w e l l beyond the optimal f o c a l range f o r c e l l d e t e c t i o n (-40 to +25 fim). Nevertheless, the o v e r a l l performance of the a l g o r i t h m was r a t h e r poor outside t h i s range, s i n c e few c e l l s were c o r r e c t l y c l a s s i f i e d i n t h i s r e g i o n . However, scanning was u n l i k e l y to take place a t these suboptimal focus l e v e l s , s i n c e f o c u s i n g algorithms to maintain focus s e t t i n g s w i t h i n the optimal range f o r c e l l d e t e c t i o n had already been developed. 5.7 P r a c t i c a l Considerations While the r e c o g n i t i o n a l g o r i t h m f o r V79 c e l l s was found to perform w i t h f a l s e p o s i t i v e e r r o r r a t e s comparable to those of a human observer, c e r t a i n precautions had to be taken to ensure that t h i s accuracy was maintained. Most pr e v a l e n t among these was the c o n t r o l of the p r o p o r t i o n of d e b r i s detected i n the t i s s u e c u l t u r e f l a s k , s i n c e the f a l s e p o s i t i v e e r r o r i s dependent on the r e l a t i v e numbers of c e l l s and n o n - c e l l s detected. Based on r e s u l t s from d i f f e r e n t f l a s k s c o n t a i n i n g v a r y i n g amounts of d e b r i s , f a l s e p o s i t i v e e r r o r s could be expected to remain below 5% i f the p r o p o r t i o n of debris d i d not exceed 30-35% of the t o t a l o b j e c t s detected. I n p r a c t i c e , t h i s l i m i t was found to be q u i t e reasonable, and those f l a s k s c o n t a i n i n g excessive amounts of debris could u s u a l l y be e a s i l y i d e n t i f i e d and r e v i s i t e d manually f o r v e r i f i c a t i o n of the c l a s s i f i c a t i o n s made by the r e c o g n i t i o n algorithm. 125 A second p r a c t i c a l c o n s i d e r a t i o n concerned the c l a s s i f i c a t i o n of c e l l doublets. Even a f t e r a l l the steps i n v o l v e d i n the c l a s s i f i c a t i o n of detected o b j e c t s were invoked (namely, l a b e l i n g of ob j e c t s w i t h m u l t i p l e peaks, a p p l i c a t i o n of the r e c o g n i t i o n a l g o r i t h m , and e x c l u s i o n of near neighbours), approximately 6% of those objects c l a s s i f i e d as c e l l s would i n f a c t be doublets. Because of the s i z e or p r o x i m i t y of t h e i r c o n s t i t u e n t c e l l s , these doublets were not detected as m u l t i p l e peaks, and th e r e f o r e produced s i g n a l s i n d i s t i n g u i s h a b l e from those of s i n g l e c e l l s . The importance of t h e i r c o n t r i b u t i o n to the r e s u l t s of a given type of experiment t h e r e f o r e had to be evaluated before the r e c o g n i t i o n a l g o r i t h m c o u l d be r e l i e d upon e n t i r e l y f o r the i d e n t i f i c a t i o n of c e l l s . S i m i l a r l y , the p o s s i b i l i t y that c e l l s m i s c l a s s i f i e d as d e b r i s by the r e c o g n i t i o n a l g o r i t h m represented a subpopulation w i t h a s i g n i f i c a n t l y d i f f e r e n t response to the treatment of i n t e r e s t than the c o r r e c t l y c l a s s i f i e d c e l l s had to be i n v e s t i g a t e d . However, when comparisons of s u r v i v a l data from populations w i t h and without doublets were made, i n c l u s i o n of the doublets d i d not appear to have s i g n i f i c a n t e f f e c t s on the r e s u l t s . S i m i l a r f i n d i n g s r e s u l t e d when s u r v i v a l data from populations t h a t contained e i t h e r manually or a u t o m a t i c a l l y c l a s s i f i e d objects were compared. 126 6. MEASUREMENT OF CELL GROWTH AND COLONY SIZE 6.1 Preliminary Considerations A primary goal i n the development of methods to determine c e l l colony s i z e s f o r s u r v i v a l estimates was to make the measurements as r a p i d l y as p o s s i b l e , s i n c e l a r g e - s c a l e experiments i n v o l v i n g s e v e r a l f l a s k s were to be performed. The use of a minimal amount of in f o r m a t i o n to generate the de s i r e d r e s u l t s was there f o r e emphasized. As to the accuracy w i t h which colony s i z e estimates had to be made, the goal was to d i s t i n g u i s h c o l o n i e s w i t h more than 50 c e l l s from c o l o n i e s w i t h l e s s than 50 c e l l s , s i n c e t h i s i s the commonly used c r i t e r i o n by which s u r v i v o r s are d i s t i n g u i s h e d from non-survivors. This standard, while considered to be somewhat a r b i t r a r y (Nias e t a l , 1965; A l p e r , 1979), i s based on the observations of Puck and Marcus (1956), who, i n the generation of the f i r s t mammalian c e l l s u r v i v a l curves, found that l e t h a l l y i r r a d i a t e d c e l l s may undergo up to 5 or 6 d i v i s i o n s before l o s i n g t h e i r p r o l i f e r a t i v e c a p a c i t y . Even w i t h a 5 0 - c e l l l i m i t on non-s u r v i v i n g colony s i z e , i n c u b a t i o n times s u f f i c i e n t l y long to all o w a c l e a r d i s t i n c t i o n between s u r v i v o r s and non-survivors are g e n e r a l l y not p r a c t i c a b l e . Frequently, s u r v i v o r s growing at normal r a t e s obscure slow-growing and no n - s u r v i v i n g clones before they can be un e q u i v o c a l l y d i s t i n g u i s h e d from each other. With V79 c e l l s , f o r example, the growth r a t e f o r the slowest-growing s u r v i v o r s i n an i r r a d i a t e d p o p u l a t i o n has been estimated to be up to 3 times longer ( i n terms of po p u l a t i o n doubling time) than the normal growth r a t e ( S i n c l a i r , 1964). N o r m a l l y - p r o l i f e r a t i n g s u r v i v i n g c o l o n i e s would t h e r e f o r e c o n t a i n over 10 5 c e l l s before the slowest-growing s u r v i v o r s reached the 5 0 - c e l l mark. Such 127 c o n d i t i o n s are d i f f i c u l t to accommodate, e s p e c i a l l y i n low dose s t u d i e s where the number of normally p r o l i f e r a t i n g s u r v i v o r s i s l a r g e . Since i n c u b a t i o n times s u f f i c i e n t l y long to produce a c l e a r d i s t i n c t i o n between s u r v i v o r s and non-survivors are not p r a c t i c a l , the 5 0 - c e l l c r i t e r i o n i s g e n e r a l l y a p p l i e d a f t e r a s p e c i f i e d i n c u b a t i o n time th a t i s kept constant from experiment to experiment. Consequently, r e l a t i v e l y s u b t l e d i f f e r e n c e s i n colony s i z e have to be detected. Furthermore, c o n t r o l l i n g the " f a l s e negative" e r r o r i s j u s t as important as c o n t r o l l i n g the " f a l s e p o s i t i v e " e r r o r , so there i s l i t t l e leeway f o r manoeuvering d e c i s i o n boundaries i n a c l a s s i f i c a t i o n a l g o r i t h m . 6.2 Morphological Features of Growing C e l l Populations The morphology of a growing c e l l p o p u l a t i o n can be q u i t e d i f f e r e n t from that of f r e s h l y tryps i n i z e d and p l a t e d c e l l s . For V79 and CHO c e l l s , the two daughter c e l l s a r i s i n g from d i v i s i o n of a s i n g l e parent c e l l tend to remain i n c l o s e p r o x i m i t y . Thus, a f t e r s e v e r a l generations, a s i n g l e c e l l gives r i s e to a colony of closely-packed progeny. Examples of the appearance of both V79 and CHO c o l o n i e s are shown i n Figure 39. V79 c o l o n i e s are more compact than CHO c o l o n i e s . CHO c e l l s tend to s c a t t e r near colony edges, sometimes making i t d i f f i c u l t to d e f i n e the colony boundaries. V79 c o l o n i e s have reasonably d i s t i n c t boundaries, but i t i s d i f f i c u l t to d i s t i n g u i s h i n d i v i d u a l c e l l s w i t h i n them. The morphology of a growing c e l l p o p u l a t i o n can be a f f e c t e d by the age of the c u l t u r e . V79 and CHO c e l l s , f o r example, tend to cease growing as a monolayer once a colony reaches a c e r t a i n s i z e . D i v i d i n g c e l l s near the colony center s t a r t to p i l e up on top of each other. This can be seen i n the o l d e r c o l o n i e s of Figures 39b,d. 128 a .-'•••6-'o /.v- '..;<*:., ;.-'•; .V'.;.. • . d Figure 39. V 7 9 and CHO c o l o n i e s a t 3 days and 5 days a f t e r p l a t i n g , (a) V 7 9 , 3-day-old, (b) V 7 9 , 5 - d a y - o l d , ( c ) CHO, 3-day-old, (d) CHO, 5 - d a y - o l d . 129 The morphology of a c e l l p o p u l a t i o n can a l s o be a f f e c t e d by environmental c o n d i t i o n s other than those r e l a t e d to p o p u l a t i o n d e n s i t y . Factors such as temperature or oxygen and n u t r i e n t supply can a f f e c t c e l l shape, and t h e r e f o r e the o v e r a l l appearance of the colony. Figure 40 shows the e f f e c t o f temperature changes on V79 c e l l morphology, as w e l l as the e f f e c t of oxygen and n u t r i e n t d e p r i v a t i o n on the appearance of CHO c e l l s . V79 c e l l s are f l a t t e n e d and elongated at 37 °C, but begin to round up when l e f t at room temperature. In the other example shown, CHO c e l l s are seen to elongate, f l a t t e n , and disperse when deprived of oxygen and n u t r i e n t s . Other environmental f a c t o r s that can a f f e c t c e l l morphology i n c l u d e the type and q u a n t i t y of serum i n the growth medium (Puck et a l , 1956), as serum provides c e l l s w i t h growth, m o t i l i t y , and adhesion f a c t o r s (Armelin and Armelin, 1978). C e l l u l a r damage, which may a r i s e through spontaneous mutation or through the a c t i o n of damaging agents, can a l s o a f f e c t c e l l morphology (Figure 40e) . S o - c a l l e d g i a n t c e l l s , f o r example, are commonly seen i n r a d i a t i o n t r e a t e d populations (Puck et a l , 1957; R o n d a n e l l i et a l , 1966), and can be found w i t h low frequency i n u n i r r a d i a t e d samples as w e l l . Colonies made up of abnormally s m a l l , densely packed c e l l s can a l s o occur. A range of more s u b t l e d i f f e r e n c e s , many of which simply r e f l e c t the v a r i a b i l i t y inherent i n the p o p u l a t i o n , a l s o e x i s t s . 1 3 0 v#? a OCT c o § 0 0 /, C ° o o H b c d Figure 40. Some effects o f environmental conditions and damaging agents on c e l l morphology. (a) CHO c e l l s , normal conditions, (b) CHO c e l l s , oxygen and nutrient deprived, (c) V79 c e l l s , normal conditions, (d) V79 c e l l s , after 5 hr at room temperature, (e) V79 c e l l s , radiation damaged. 131 6.3 S i g n a l Features of C e l l Colonies Because i t was important to perform measurements on c e l l c o l o n i e s w i t h reasonable speed, colony scanning was performed at the same low r e s o l u t i o n (1.3 pm p i x e l s i z e at 10X m a g n i f i c a t i o n ) used i n the c e l l d e t e c t i o n and r e c o g n i t i o n procedures. T y p i c a l l i n e scans across V79 and CHO c o l o n i e s at t h i s r e s o l u t i o n are shown i n Figure 41 f o r c o l o n i e s of two d i f f e r e n t ages (3 days a f t e r p l a t i n g and 5 days a f t e r p l a t i n g ) . In c o n t r a s t to a scan across the center of an i n d i v i d u a l c e l l , l i n e scans across c o l o n i e s d i d not have a uniquely c h a r a c t e r i z a b l e s t r u c t u r e . M u l t i p l e peaks and troughs occurred i n the s i g n a l , but these d i d not n e c e s s a r i l y correspond i n a s p e c i f i c way to i n d i v i d u a l c e l l s , e s p e c i a l l y i n the f l a t t e n e d , c l o s e l y packed V79 c o l o n i e s . Furthermore, the "heaping up" of c e l l s i n o l d e r c o l o n i e s caused i d e n t i f i a b l e changes i n the colony l i n e scans, i n c l u d i n g a decrease i n the s i g n a l i n t e n s i t y . Although a r e g u l a r s t r u c t u r e r e l a t i n g d i r e c t l y to i n d i v i d u a l c e l l s w i t h i n a colony was d i f f i c u l t to d i s c e r n i n a s i n g l e l i n e scan, s e v e r a l q u a n t i t a t i v e colony s i g n a l features were c a l c u l a t e d u s i n g the SSCAN program. Among these features were the number of peaks above and the number of troughs below s e l e c t e d t h r e s h o l d l e v e l s . The thresholds used were 20 /Mm above background f o r peak measurements and 20 um below background f o r trough measurements. A d d i t i o n a l f e a t u r e s were al s o c a l c u l a t e d . These are shown, along w i t h the names used to i d e n t i f y them, i n Figure 42. Feature values were summed over s e v e r a l e q u a l l y spaced l i n e scans to generate a feature score. In t h i s way, features such as "area below" become two-dimensional, even though they appear one-dimensional f o r the s i n g l e l i n e scans i l l u s t r a t e d i n Figure 42. A s i m i l a r argument f o l l o w s f o r the "volume" f e a t u r e s . 132 V79 day 3 CHO day 3 V79 day 5 CHO day 5 q200 ^150 -^100 - 0 grey level intensity scale F i g u r e 41. L i n e s c a n s a c r o s s V79 and CHO c e l l c o l o n i e s ( i n d i v i d u a l p i x e l p o i n t s d e t e c t e d by the l i n e a r s e n s o r have been j o i n e d by s t r a i g h t l i n e s ) . 133 area above - area below = number of p i x e l s above number of p i x e l s below s p e c i f i e d t h r e s h o l d s p e c i f i e d t h r e s h o l d S ( p i x e l i n t e n s i t y - t h r e s h o l d i n t e n s i t y ) E ( t h r e s h o l d i n t e n s i t y - p i x e l i n t e n s i t y ) f o r p i x e l s above s p e c i f i e d t h r e s h o l d f o r p i x e l s below s p e c i f i e d t h r e s h o l d t o t a l area = t o t a l volume = area above + area below volume above + volume below Figure 42. S i x of the SSCAN colony feat u r e s . I n d i v i d u a l p i x e l s are shown on the sample l i n e scans. Thresholds were set at 10 grey l e v e l s above and below background i n t e n s i t y . 134 6.4 E s t i m a t i o n of Colony S i z e and Popu l a t i o n Density 6.4.1 C o l l e c t i o n of Feature Data P o p u l a t i o n d e n s i t y data was gathered u s i n g the SSCAN program. In order to provide r a p i d scanning, l i n e scans were taken at 4 /im i n t e r v a l s across p r e v i o u s l y s e l e c t e d regions of i n t e r e s t i n the f l a s k . Only the middle 256 p i x e l s of the s i g n a l detected by the l i n e a r sensor were analyzed i n any given scan. A t o t a l of 64 such l i n e scans made up each colony scan, y i e l d i n g a 256 /xm X 333 /im scanning area (the y-dimension was l a r g e r than the x-dimension because of the 1.3 /im p i x e l width i n t h i s d i r e c t i o n ) . This area c o u l d be scanned i n approximately 1 second, and was centered by the SSCAN program on a p r e v i o u s l y s e l e c t e d set of x,y coord i n a t e s . These coordinates u s u a l l y marked the l o c a t i o n of a parent c e l l found by the c e l l d e t e c t i o n procedure on the day of p l a t i n g . The scanning area was la r g e enough to encompass c o l o n i e s c o n s i s t i n g of s e v e r a l hundred c e l l s . Accurate s i z e determinations of c o l o n i e s l a r g e r than t h i s were not r e q u i r e d i f the 5 0 - c e l l c r i t e r i o n was to be used to measure c e l l s u r v i v a l . 6.4.2 C o r r e l a t i o n of Feature Scores w i t h P o p u l a t i o n Density The most s t r a i g h t f o r w a r d way to determine how w e l l colony feature scores corresponded to the a c t u a l number of c e l l s present i n the region being scanned was to compare manual counts w i t h data obtained by the C e l l Analyzer. However, t h i s method was extremely d i f f i c u l t to c a r r y out f o r c o l o n i e s l a r g e r than approximately 50-100 c e l l s . For V79 c e l l s , the d i f f i c u l t i e s were compounded by the c l o s e a s s o c i a t i o n of the f l a t t e n e d c e l l s , which made them extremely hard to d i s t i n g u i s h from each other even i n c o l o n i e s c o n t a i n i n g as few as 30 c e l l s . An a l t e r n a t i v e , a l b e i t l e s s p r e c i s e , method to measure the r e l a t i o n s h i p between the number of c e l l s i n 135 the scanning area and the measured colony f e a t u r e scores was t h e r e f o r e employed. This method took advantage of the f a c t t h a t c e l l s p l a t e d i n t o a t i s s u e c u l t u r e f l a s k are randomly d i s t r i b u t e d over the f l a s k s u r f ace. The c e l l d e n s i t y at one l o c a t i o n could t h e r e f o r e be expected to be the same as the c e l l d e n s i t y i n another l o c a t i o n , even a f t e r colony formation, provided that the regions scanned were not too s m a l l . C e l l d e n s i t y i n a s e l e c t e d r e g i o n would be r e p r e s e n t a t i v e of the o v e r a l l c e l l d e n s i t y i n the e n t i r e f l a s k . The l a t t e r q u a n t i t y could be measured w i t h r e l a t i v e ease by t r y p s i n i z i n g the c e l l s and determining the c o n c e n t r a t i o n of the r e s u l t i n g suspension w i t h a C o u l t e r Counter. Colony fe a t u r e scores f o r a s e l e c t e d r e g i o n c o u l d be compared to the o v e r a l l d e n s i t y estimated from the C o u l t e r counts to give an i n d i r e c t measure of the r e l a t i o n s h i p between colony fe a t u r e values and the number of c e l l s present i n the r e g i o n being scanned. Using the above method, estimates of the c o r r e l a t i o n between feature scores and the number of c e l l s present were made f o r c o l o n i e s of v a r i o u s ages. Data f o r V79 c e l l s p l a t e d at two d i f f e r e n t i n i t i a l d e n s i t i e s are shown i n Figure 43a f o r the " t o t a l area" colony f e a t u r e . Feature scores were obtained f o r a 92 mm2 area w i t h i n the f l a s k . Other colony features behaved i n a s i m i l a r f a s h i o n to the " t o t a l area", w i t h p o p u l a t i o n growth r a t e s estimated from the colony features c o n s i s t e n t l y underestimating the a c t u a l growth r a t e of the p o p u l a t i o n . Because of the i n c r e a s i n g compactness of the V79 c o l o n i e s as they grew, the mean fe a t u r e score per c e l l decreased s t e a d i l y as the p o p u l a t i o n aged. Figure 43b i l l u s t r a t e s these changes f o r the " t o t a l area" f e a t u r e . Again, the other colony features showed a s i m i l a r dependence on colony age. For the p l o t of Figure 43b, data were obtained from samples w i t h s e v e r a l d i f f e r e n t i n i t i a l p l a t i n g d e n s i t i e s . As long as there was s u f f i c i e n t room f o r the c o l o n i e s 136 IO 4 CO 'a; o CO CD o a * ce CO CC o (a) 1 0 3^ t d = 9 . 0 h r 10 2 , 10U 10° Q'"* t d =13 .9 h r - • • A - c e l l dens i t y • o A - " t o t a l a r e a " - i 1 . 1 . 1 . 1 1 1 1 1 r 0 20 40 60 80 100 120 140 CO 10 5 % 10 4 R 10 3 CO <D u CO CO -+-> o 10 2?" i — | — . — | — i — | — i — | — i — | — i — | — i -0 20 40 60 80 100 120 140 t i m e s i n c e p l a t i n g (h r ) Figure 43. Measurement of V79 population density using colony feature scores. (a) c e l l density (as measured by t r y p s i n i z a t i o n and counting with Coulter counter) compared to " t o t a l area" per mm2 (measurements were made i n common f l a s k s ) . (b) " t o t a l area" per c e l l as a function of colony age. 137 to grow, the feat u r e score per c e l l d i d not appear to depend on the i n i t i a l p l a t i n g d e n s i t y . CHO c e l l c o l o n i e s , on the other hand, d i d not e x h i b i t a continuous change i n the mean feature score per c e l l as the c e l l p o p u l a t i o n aged, at l e a s t not u n t i l the c e l l s s t a r t e d p i l i n g up at 5-6 days p o s t - p l a t i n g . Indeed, the rounded, d i s c r e t e c e l l s i n CHO c o l o n i e s allowed a one-to-one correspondence to be obtained between the number of peaks above t h r e s h o l d ( s u b j e c t to minor r e s t r a i n t s ) and the number of c e l l s present (Figure 44). The r e s t r a i n t s p l a c e d on the counting of peaks were as f o l l o w s : (1) only peaks w i t h widths between 3 and 14 p i x e l s above the d e t e c t i o n t h r e s h o l d were counted, and (2) i f peaks occurred w i t h i n 4 m^ of each other on 2 adjacent l i n e scans, they were considered as belonging to the same c e l l . In a d d i t i o n , because CHO c e l l s were on average somewhat l a r g e r than V79 c e l l s , a 6 /im step was taken between scans i n order to minimize the occurrence of m u l t i p l e h i t s on s i n g l e c e l l s , even though there was a s l i g h t r i s k of missing some of the smaller c e l l s i n the p o p u l a t i o n . Improving the accuracy of po p u l a t i o n d e n s i t y estimates f o r V79 c e l l s would r e q u i r e a more d e t a i l e d a n a l y s i s of the e f f e c t of colony s i z e and age on the r e l a t i o n s h i p between feature scores and c e l l number to be c a r r i e d out. In a d d i t i o n , measurements would l i k e l y have to be taken at higher r e s o l u t i o n , r e q u i r i n g longer scanning times. Even without such improvements, however, feature scores c a l c u l a t e d by SSCAN were p o t e n t i a l l y u s e f u l f o r d i s t i n g u i s h i n g s u r v i v i n g from n o n - s u r v i v i n g c o l o n i e s . Such c l a s s i f i c a t i o n s are g e n e r a l l y c a r r i e d out under w e l l - s p e c i f i e d c o n d i t i o n s of colony age, and, furthermore, do not r e q u i r e accurate s i z e estimates f o r most of the c o l o n i e s i n a f l a s k . Attempts were t h e r e f o r e made to use the colony f e a t u r e scores generated by the SSCAN program f o r t h i s purpose. 138 ioS O 20 40 60 80 100 120 time since plating (hr) Figure 44. Measurements of CHO population density by counting peaks i n l i n e scans spaced 6 /im apart. Results are compared to population densities measured using t r y p s i n i z a t i o n and counting with the Coulter counter. 139 6.5 Use of V79 Colony Size Information to Distinguish Surviving Colonies  from Non-surviving Colonies 6.5.1 Optimization of Factors Affecting Sample Size Since the accuracy of the data obtained i n a c e l l s u r v i v a l experiment i s dependent on the number of c e l l s assayed, f a c t o r s a f f e c t i n g t h i s number had to be i d e n t i f i e d and, i f p o s s i b l e , optimized. Among these f a c t o r s were the performance of the c e l l d e t e c t i o n and r e c o g n i t i o n procedures, considered p r e v i o u s l y i n Chapters 4 and 5. Even w i t h optimal procedures f o r c e l l d e t e c t i o n and r e c o g n i t i o n , however, the number of c e l l s t h a t could be used f o r s u r v i v a l measurements was l i m i t e d by s p a t i a l c o n s t r a i n t s . E x c l u s i o n of near neighbours from the data set was necessary to prevent non-survivors and slow-growing s u r v i v o r s from being obscured by l a r g e neighbouring c o l o n i e s before t h e i r s u r v i v a l s t a t u s c o u l d be determined. In the case of automated colony s c o r i n g , even r e l a t i v e l y minor cases of i n t e r f e r e n c e would be d i f f i c u l t to deal w i t h , s i n c e the shape and arrangement of c o l o n i e s w i t h i n the scanning area were not used by the SSCAN program to determine whether there was more than one colony i n the scanning f i e l d . The e f f e c t of the near neighbour e x c l u s i o n procedure on f i n a l sample s i z e c o u l d be determined a n a l y t i c a l l y . C e l l s s e t t l e onto the f l a s k surface i n a random f a s h i o n when they are p l a t e d from a suspension, so the number of c e l l s f a l l i n g i n t o an area of given s i z e f o l l o w s a Poisson d i s t r i b u t i o n . For an unbounded p l a t i n g area, the p r o b a b i l i t y that no c e l l s f a l l i n t o an e x c l u s i o n zone of s i z e A surrounding a randomly s e l e c t e d l o c a t i o n i s : P = e'XA ( 6 . 1 ) 140 where A i s the c e l l d e n s i t y . I f a square e x c l u s i o n zone of area A = 4 5 2 i s defined, the p r o b a b i l i t y that no c e l l s f a l l i n t o i t at a given l o c a t i o n becomes: The CSCAN program u t i l i z e s such a square e x c l u s i o n r e g i o n , w i t h S being d e f i n e d as the e x c l u s i o n d i s t a n c e . The p r o b a b i l i t y t h a t no other c e l l s f a l l i n t o the e x c l u s i o n zone around a given c e l l (the l o c a t i o n of which i s equi v a l e n t to any randomly s e l e c t e d l o c a t i o n i n the f l a s k ) i s then P(S). Ignoring d e b r i s , i f N i s the t o t a l number of c e l l s p l a t e d i n t o a 25 cm 2 f l a s k , the number of c e l l s , n, expected to be in c l u d e d i n the data set a f t e r e x c l u s i o n i s : where 5 i s measured i n mm, 0.0004JV i s the c e l l d e n s i t y A i n cells/mm 2, and the f a c t o r 0.4 a r i s e s because the 10 cm 2 area scanned by the CSCAN program makes up only 40% of the t o t a l area onto which the N c e l l s are p l a t e d . The scanning area has been assumed to be large enough to approximate an i n f i n i t e l y l a r g e p l a t i n g area, w i t h edge e f f e c t s making a n e g l i g i b l e c o n t r i b u t i o n to the r e s u l t s of the e x c l u s i o n procedure. According to the formula f o r n(N,6), the number of c e l l s remaining a f t e r the a p p l i c a t i o n of a given e x c l u s i o n distance can be maximized i f an optimum number of c e l l s , N , i s p l a t e d . Maximizing n(N,8) f o r f i x e d 5 y i e l d s : ?(5) = exp(-4A5 2). (6.2) n(N,8) = 0.4W P(6) - 0AN exp(-0.0016N5 2) (6.3) n opt N - 1 / 0.00165 2 opt ' t = 2505" 2e" 1 = 91.975 -2 (6.4) (6.5) 141 P l o t s of n(N,8) versus N and n Q p t versus 6 ave shown i n Figure 45. When 6 exceeds approximately 0.5 mm, few c e l l s would be expected to remain a f t e r e x c l u s i o n . Furthermore, not a l l objects considered by the CSCAN e x c l u s i o n procedure were c e l l s . In general, 20-30% of the detected o b j e c t s were d e b r i s , which was in c l u d e d i n the e x c l u s i o n c a l c u l a t i o n s because a c e r t a i n proportion- of c e l l s was m i s c l a s s i f i e d by the r e c o g n i t i o n a l g o r i t h m . The p r o x i m i t y of these c e l l s to other c e l l s i n the data s et had to be considered, f o r c i n g a l l objects c l a s s i f i e d as n o n - c e l l s to be considered along w i t h them. I f the detected debris was in c l u d e d i n the data s e t , e x c l u s i o n distances no l a r g e r than 0.35-0.40 mm cou l d be used i f data sets of at l e a s t 400-500 c e l l s per f l a s k (as suggested i n the I n t r o d u c t i o n , S e c t i o n 1.1.3) were to be obtained. With n o n - c e l l o b j e c t s i n c l u d e d i n N, the optimal number of c e l l s p l a t e d per f l a s k f o r these e x c l u s i o n distances would be 2500-3500. 6.5.2 C l a s s i f i c a t i o n of Colonies Using a S i n g l e Colony S i z e Measurement The goal of an automated colony c l a s s i f i c a t i o n procedure was to provide s u r v i v a l estimates comparable to those obtained through manual s c o r i n g , so the f i r s t step i n the development of the a l g o r i t h m was to determine the c o n d i t i o n s under which manual c l a s s i f i c a t i o n c o u l d be performed. V79 c o l o n i e s are commonly c l a s s i f i e d manually a f t e r a 6-7 day in c u b a t i o n p e r i o d , both i n our, and other, l a b o r a t o r i e s (e.g. G i l l e s p i e e t a l , 1975b; Hoshi et a l , 1988). This i n c u b a t i o n p e r i o d i s s u f f i c i e n t l y long to produce macroscopic c o l o n i e s that can be e a s i l y scored i n a t y p i c a l high dose experiment. However, because c o l o n i e s were examined m i c r o s c o p i c a l l y whenever the C e l l Analyzer was used, a l a r g e p r o p o r t i o n of them (70-90%) could be c l a s s i f i e d as e a r l y as 4 days p o s t - p l a t i n g . These c o l o n i e s c o n s i s t e d p r i m a r i l y of those that already had more than 50 c e l l s , 1 4 2 800 6 (mm) F i g u r e 45. E f f e c t o f p r o x i m i t y e x c l u s i o n on s i z e o f d a t a s e t . (a) number o f c e l l s l e f t i n the 10 cm 2 s c a n n i n g a r e a a f t e r p r o x i m i t y e x c l u s i o n v e r s u s number o f c e l l s p l a t e d i n t o the 25 cm 2 t i s s u e c u l t u r e f l a s k , f o r v a r i o u s e x c l u s i o n d i s t a n c e s . (b) maximum p o s s i b l e number o f c e l l s i n the 10 cm 2 s c a n n i n g a r e a a f t e r p r o x i m i t y e x c l u s i o n , as a f u n c t i o n o f e x c l u s i o n d i s t a n c e . 143 and so cou l d be c l a s s i f i e d as s u r v i v o r s . P r e l i m i n a r y manual assessments were f r e q u e n t l y made a f t e r a 4-day i n c u b a t i o n p e r i o d , s i n c e the f l a s k was l e s s crowded and few, i f any, c o l o n i e s had grown beyond the l i m i t s of the 0.35-0.40 mm e x c l u s i o n zones necessary to o b t a i n data s e t s of reasonable s i z e . Those c o l o n i e s that c o u l d not be c l a s s i f i e d a t 4 days p o s t - p l a t i n g were r e v i s i t e d a t 6 or 7 days p o s t - p l a t i n g f o r a f i n a l assessment of t h e i r s u r v i v a l s t a t u s . Most of these c o l o n i e s could s t i l l be manually c l a s s i f i e d at t h i s time, d e s p i t e the presence of some la r g e s u r v i v i n g c o l o n i e s . While manual c l a s s i f i c a t i o n s could be made without too much d i f f i c u l t y a f t e r a 6-day i n c u b a t i o n p e r i o d , the use of an automated c l a s s i f i c a t i o n procedure at t h i s time was not f e a s i b l e unless u n s u i t a b l y l a r g e (0.6-1.0 mm) e x c l u s i o n distances were used. However, si n c e i t had been found that 70-90% of c o l o n i e s could be manually scored w i t h confidence at 4 days p o s t - p l a t i n g , the p o s s i b i l i t y that a comparable p r o p o r t i o n could be c l a s s i f i e d u s i n g the SSCAN colony features a f t e r a s i m i l a r i n c u b a t i o n p e r i o d was i n v e s t i g a t e d . Sample data f o r both i r r a d i a t e d and u n i r r a d i a t e d c e l l s were c o l l e c t e d . C e l l s were i r r a d i a t e d i n suspension w i t h 0-3 Gy of 250 kVp X-rays immediately before p l a t i n g them i n t o the Nunclon f l a s k s . This y i e l d e d s u r v i v i n g f r a c t i o n s of 0.3-0.9. Colony fea t u r e data were c o l l e c t e d d a i l y from 1 day to 5 days p o s t - i r r a d i a t i o n . These feature scores were then compared to manual c l a s s i f i c a t i o n s i n order to determine whether they could be used to d i s t i n g u i s h s u r v i v i n g from non - s u r v i v i n g c o l o n i e s . A comparison of manual colony c l a s s i f i c a t i o n s (where f i n a l c l a s s i f i c a t i o n s were decided on day 6) w i t h the " t o t a l area" f e a t u r e score at 4 days p o s t - i r r a d i a t i o n i s shown i n the histogram of Figure 46. This histogram represents combined data f o r a l l dose p o i n t s from 4 experiments. E v i d e n t l y , there i s a re g i o n of overlap where some c o l o n i e s manually 0.0 0.2 0.4 0.6 0.8 1.0 total area (pixels x 104) Figure 46. Day 4 " t o t a l area" feature scores f o r s u r v i v i n g and non-su r v i v i n g V79 c o l o n i e s a r i s i n g from an i r r a d i a t e d (0-2.4 Gy of 250 kVp X-rays) parent c e l l p o p u l a t i o n . The c r i t e r i o n f o r s u r v i v a l was a minimum of 50 c e l l s per colony by 6 days post-treatment. 145 c l a s s i f i e d as s u r v i v o r s have a lower feature score than some c o l o n i e s c l a s s i f i e d as non-survivors. Nevertheless, i t was p o s s i b l e to s e l e c t a d e c i s i o n boundary f o r which the number of " f a l s e " s u r v i v o r s e q u a l l e d the number of " f a l s e " non-survivors, thereby g i v i n g a c o r r e c t s u r v i v a l estimate. Such a d e c i s i o n boundary could be l o c a t e d f o r colony s i z e data obtained as e a r l y as two days p o s t - i r r a d i a t i o n , but the degree of overlap between s u r v i v o r s and non-survivors was s u b s t a n t i a l f o r i n c u b a t i o n periods l e s s than 4 days. By day 4, the number of over l a p p i n g c o l o n i e s was g e n e r a l l y only 2-6% of the sample s i z e f o r the " t o t a l area" f e a t u r e (on day 3 i t was 6-10% and on day 2 i t was about 20%). Furthermore, there was no s i g n i f i c a n t improvement i n the degree of overlap between day 4 and 5. In f a c t , sometimes the overlap was greater on day 5, probably as a r e s u l t of the extension of some la r g e c o l o n i e s i n t o neighbouring e x c l u s i o n regions. This occurred even when e x c l u s i o n d i s t a n c e s as l a r g e as 0.6 mm were used. Day 4 c o l o n i e s , on the other hand, were f o r the most p a r t completely contained even i n 5=0.35 mm e x c l u s i o n areas. Even though a d e c i s i o n boundary w i t h r e l a t i v e l y l i t t l e o verlap could be i d e n t i f i e d a t 4 days p o s t - p l a t i n g , the use of such a s i n g l e t h r e s h o l d proved u n r e l i a b l e s i n c e the number of s u r v i v o r s r e l a t i v e to the number of non-survivors changes w i t h dose. The l o c a t i o n of the d e c i s i o n t h r e s h o l d would be expected to s h i f t towards higher values as the s u r v i v i n g f r a c t i o n decreases, and t h i s was indeed observed exp e r i m e n t a l l y . C o r r e c t i o n s f o r such behavior could only be made w i t h a priori knowledge of the c e l l u l a r response to the treatment of i n t e r e s t . In a d d i t i o n , the l o c a t i o n of the d e c i s i o n t h r e s h o l d was found to vary from experiment to experiment. I t was th e r e f o r e d i f f i c u l t to define a t h r e s h o l d that produced c o n s i s t e n t l y good s u r v i v a l estimates from day to day and from dose p o i n t to dose p o i n t . 1 4 6 The p o s s i b i l i t y of f i n d i n g the d e c i s i o n t h r e s h o l d f o r a p a r t i c u l a r dose p o i n t on the b a s i s of the d i s t r i b u t i o n of f e a t u r e scores was considered, s i n c e i t was conceivable that a bimodal d i s t r i b u t i o n f o r s u r v i v o r s and non-survivors would be i d e n t i f i a b l e i n the low dose range. However, such a d i s t r i b u t i o n was not evident at doses higher than approximately 1 Gy (Figure 47). These observations concur w i t h manually obtained data from Nias et a l (1965), who could f i n d no d i s t i n c t bimodal d i s t r i b u t i o n i n colony s i z e f o r i r r a d i a t e d HeLa c e l l s , even as long as 14 days a f t e r treatment. A d e c i s i o n t h r e s h o l d would t h e r e f o r e be d i f f i c u l t to i d e n t i f y f o r the higher doses on the b a s i s of the colony s i z e d i s t r i b u t i o n alone. Even where a bimodal d i s t r i b u t i o n i s i d e n t i f i a b l e , as i n Figure 47c, the l o c a t i o n of the c o r r e c t d e c i s i o n t h r e s h o l d would not n e c e s s a r i l y be a s t r a i g h t f o r w a r d procedure, s i n c e methods of s p l i t t i n g such d i s t r i b u t i o n s (e.g. V i d a l et a l , 1973) may not y i e l d d e c i s i o n boundaries that correspond to the t h r e s h o l d between s u r v i v o r s and non-survivors. Because of the d i f f i c u l t y i n i d e n t i f y i n g a s i n g l e d e c i s i o n t h r e s h o l d to c l a s s i f y c o l o n i e s , a d e c i s i o n i n t e r v a l was def i n e d i n s t e a d . This i n t e r v a l c o u l d be used to "screen" each sample. Colonies w i t h feature scores below the i n t e r v a l were c l a s s i f i e d as non-survivors, w h i l e c o l o n i e s w i t h f e a t u r e scores above the i n t e r v a l were c l a s s i f i e d as s u r v i v o r s . A l l c o l o n i e s w i t h feature scores f a l l i n g w i t h i n the i n t e r v a l had to be manually c l a s s i f i e d . The screening i n t e r v a l could be s e l e c t e d so as to absorb d i f f e r e n c e s i n colony s i z e d i s t r i b u t i o n s between dose p o i n t s and between experiments. For the t e s t data of Figures 46 and 47, i t was found t h a t a screening i n t e r v a l spanning 1000-2500 p i x e l s f o r the " t o t a l area" f e a t u r e at 4 days p o s t - i r r a d i a t i o n produced estimates of the s u r v i v i n g f r a c t i o n , S, that d e v i a t e d from manual estimates by only AS = ±0.008 i n the 0.3-0.9 s u r v i v a l 147 • i -H fl o "o o s fl 4004 3004 200-1 100^ 0 (a) < 1 Gy classified 4004 3004 2004 1001 0 • non-survivors • survivors (c) < 1 Gy unclassified 1 1 1 I i I i I r™ 0 -1 (b) > 1 Gy classified • non-survivors • survivors (d) > 1 Gy unclassified — i — i — i 7*""*i>-n-0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 total area (pixels x 104) Figure 47. Day 4 " t o t a l area" feature scores f o r V79 c o l o n i e s a r i s i n g from parent c e l l s i r r a d i a t e d at d i f f e r e n t doses. (a) featu r e scores f o r s u r v i v o r s and non-survivors a r i s i n g from c e l l s i r r a d i a t e d at 0-1 Gy. (b) features scores f o r s u r v i v o r s and non-survivors a r i s i n g from c e l l s i r r a d i a t e d at 1.0-2.4 Gy. (c) same c o l o n i e s as i n ( a ) , but u n c l a s s i f i e d i n t o s u r v i v o r s and non-survivors. (d) same c o l o n i e s as i n (b) , but u n c l a s s i f i e d i n t o s u r v i v o r s and non-survivors. 148 r a n g e . T h i s i s l e s s t h a n the u n c e r t a i n t y f o r t h e manual c l a s s i f i c a t i o n s t h e m s e l v e s , w h i c h was e s t i m a t e d t o be AS = ±0.02 on t h e b a s i s o f day 6 c o l o n i e s t h a t were d i f f i c u l t t o c l a s s i f y because t h e y c o n t a i n e d c l o s e t o 50 c e l l s . A p p r o x i m a t e l y 85-90% o f t h e c o l o n i e s c o u l d be c l a s s i f i e d on day 4 u s i n g t h e d e c i s i o n i n t e r v a l . The s e l e c t e d i n t e r v a l was t h e s m a l l e s t t h a t c o u l d be u s e d w i t h t he " t o t a l a r e a " f e a t u r e w i t h o u t i n t r o d u c i n g u n a c c e p t a b l y l a r g e e r r o r s f o r some samples. The " t o t a l a r e a " f e a t u r e showed t h e b e s t s c r e e n i n g p e r f o r m a n c e among t h e SSCAN c o l o n y f e a t u r e s , s i n c e i t had t h e l e a s t amount o f o v e r l a p between s u r v i v o r s and n o n - s u r v i v o r s . The " t o t a l a r e a " f e a t u r e c o u l d t h e r e f o r e be u s e d t o s i g n i f i c a n t l y r e d u c e , b u t n o t t o t a l l y e l i m i n a t e , t h e amount o f manual e f f o r t r e q u i r e d t o c l a s s i f y c o l o n i e s . N e v e r t h e l e s s , t he number o f c o l o n i e s t h a t h ad t o be m a n u a l l y c l a s s i f i e d d i d n o t g r e a t l y e x c e e d the number t h a t w o u l d have had t o be r e v i s i t e d t w i c e (on day 4 and day 6) even i f a l l c o l o n y c l a s s i f i c a t i o n s were made m a n u a l l y . The e x p e r i m e n t e r c o u l d t h e r e f o r e be sa v e d a c o n s i d e r a b l e amount o f e f f o r t , e s p e c i a l l y i f c o l o n i e s i n t h e s c r e e n i n g i n t e r v a l were m a n u a l l y r e v i s i t e d o n l y on day 6, r a t h e r t h a n on b o t h day 4 and day 6. 6.5.4 C l a s s i f i c a t i o n o f C o l o n i e s U s i n g S i z e and Growth R a t e I n f o r m a t i o n A s e c o n d method o f c l a s s i f y i n g c o l o n i e s u s i n g t h e SSCAN f e a t u r e s c o r e s was a l s o i n v e s t i g a t e d . I n t h i s method, an a t t e m p t was made t o i d e n t i f y more t h a n one f e a t u r e t h a t c o u l d be u s e d t o d i s t i n g u i s h s u r v i v o r s f rom n o n - s u r v i v o r s . T h i s c o u l d n o t be done u s i n g t h e SSCAN c o l o n y f e a t u r e s as t h e y were, s i n c e t h e y a l l e s s e n t i a l l y measure c o l o n y s i z e and t h e r e f o r e c o r r e l a t e t o o h i g h l y w i t h each o t h e r t o be u s e f u l as i n d e p e n d e n t f e a t u r e s i n a d i s c r i m i n a n t f u n c t i o n . F u r t h e r m o r e , t h e r e s o l u t i o n o f t h e c o l o n y scans was not h i g h enough to measure morphological f e a t u r e s (such as c e l l s i z e and p o p u l a t i o n d e n s i t y ) that a i d the human observer i n a s s e s s i n g the s u r v i v a l s t a t u s of c o l o n i e s c o n t a i n i n g c l o s e to 50 c e l l s (higher r e s o l u t i o n s would have r e q u i r e d longer scanning times). One independent feature that c o u l d be estimated u s i n g the SSCAN f e a t u r e scores was the growth r a t e of the c o l o n i e s , s i n c e f e a t u r e data could be c o l l e c t e d on more than one day. Growth r a t e cannot be measured e a s i l y by manual means, and could t h e r e f o r e provide i n f o r m a t i o n that would normally not be a v a i l a b l e i n the manual s c o r i n g process. Indeed, p a i n s t a k i n g manual measurements of colony growth r a t e s by Grote et a l (1981a) on i r r a d i a t e d BHK 21 C13 ( S y r i a n hamster) f i b r o b l a s t c e l l s have shown that the growth r a t e of n o n - s u r v i v i n g c o l o n i e s i s s i g n i f i c a n t l y lower than that of even the slow-growing s u r v i v o r s a f t e r approximately 2 days i n c u b a t i o n . Since the S y r i a n hamster c e l l s had a s i m i l a r doubling time to the V79 c e l l s used i n our l a b o r a t o r y , i t was p o s s i b l e t h a t an estimate of V79 colony growth r a t e a f t e r 2 days i n c u b a t i o n c o u l d be used as a feature to d i s t i n g u i s h s u r v i v o r s from non-survivors. This would, of course, r e q u i r e that two sets of colony f e a t u r e measurements be made. These were made at 2 and 4 days post-treatment, since i t had already been e s t a b l i s h e d that day 4 colony s i z e data were u s e f u l f o r d i s t i n g u i s h i n g s u r v i v o r s from non-survivors. The day 2-4 colony growth rat e and the day 4 colony s i z e c o u l d be used as independent features i n a d i s c r i m i n a n t f u n c t i o n f o r c l a s s i f y i n g c o l o n i e s . The growth r a t e of a c e l l p o p u l a t i o n i s g e n e r a l l y e x p o n e n t i a l when c o n d i t i o n s are o p t i m a l , and can be represented by the equation: N2 = N1 expiata-fcj)] (6.6) 150 where N i s the number of c e l l s present at time t , and £ i s the exponential growth constant. Colony feature scores, which give an estimate of the number of c e l l s present, could be s u b s t i t u t e d f o r N1 and N2 without changing the form of the equation. However, because the SSCAN f e a t u r e s do not c o r r e l a t e l i n e a r l y w i t h c e l l number f o r c o l o n i e s of d i f f e r e n t ages, the value c a l c u l a t e d f o r £ from the f e a t u r e scores would not be the same as the a c t u a l e x p o n e n t i a l growth constant. Nevertheless, the growth measured by the " t o t a l area" feature between day 2 and day 4 i s e s s e n t i a l l y e x p onential ( c f . Figure 43), a l l o w i n g Equation 6.6 to be a p p l i e d . V79 c e l l s exposed to 0-2.4 Gy doses of 250 kVp X-rays were used to evaluate the proposed method. Using the two f e a t u r e s (day 2-4 colony "growth r a t e " and day 4 colony s i z e ) , a d i s c r i m i n a n t f u n c t i o n to d i s t i n g u i s h s u r v i v o r s from non-survivors was generated. The d i s c r i m i n a n t f u n c t i o n was once again based on Bayes' c l a s s i f i c a t i o n r u l e (see Sections 5.4 and 10.1). U n l i k e i n S e c t i o n 5.4, however, a s i n g l e d i s c r i m i n a n t f u n c t i o n was not generated from a set of t e s t data and then a p p l i e d to a l l subsequent experiments. Rather, a separate f u n c t i o n was generated f o r each experiment on the b a s i s of a small subset of the experimental samples, u s u a l l y 1 or 2 f l a s k s . In t h i s way, any d i f f e r e n c e s i n c e l l morphology and growth r a t e that occur from experiment to experiment c o u l d be accounted f o r . Such d i f f e r e n c e s between experiments can a r i s e from day-to-day v a r i a t i o n s i n the c e l l p o p u l a t i o n , d i f f e r e n t i n c u b a t i o n periods before and between scans, and small f l u c t u a t i o n s i n environmental c o n d i t i o n s (e.g. temperature and pH). Both l i n e a r and quadratic d i s c r i m i n a n t f u n c t i o n s (see Appendix, S e c t i o n 10.1) were t e s t e d f o r t h e i r a b i l i t y to d i s t i n g u i s h s u r v i v i n g from n o n - s u r v i v i n g c o l o n i e s . An example of a quadratic s e p a r a t i o n i s shown i n Figure 48. Quadratic f u n c t i o n s were found to perform somewhat b e t t e r than 151 O X <P u cd .+-> o 0 . 0 4 1.0 survivors • non-survivors —quadra t i c decision boundary 1.0-• • • i i T i ) 111111111111111111111 I i fl i | I I 191 i i i i | i i i i | i i i i | i 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 1.0 0.8 0.6 0.4-0.2-0.0 1.2 Gy " -1 TT J T T T 1 | I I T I y T T I I J T T T T ^ T I I I j M I ' T f l T ' l ' •2.0 Gy | T I M | M T T | T T r r ] ^ T T ! | l ! PTJ"T ! T T'j I T T T 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 day 2-4 growth constant ( h r - 1 x 10~ 2) Figure 48. Separation of survivors from non-survivors using a quadratic discriminant function. The colony features measured were the " t o t a l area" at 4 days post-treatment, and the colony growth rate between 2 and 4 days post-treatment (as measured with the " t o t a l area" feature). The discriminant function was generated using the data from the 1.2 Gy dose point. 152 l i n e a r f u n c t i o n s , most l i k e l y due to the f a c t that the s u r v i v o r s and non-survivors d i d not have i d e n t i c a l feature covariance m a t r i c e s , which i s one of the c r i t e r i a f o r a p p l i c a t i o n of a l i n e a r d i s c r i m i n a n t f u n c t i o n . While t h i s c r i t e r i o n i s r a r e l y met i n most instances where a l i n e a r d i s c r i m i n a n t f u n c t i o n i s a p p l i e d , a greater number of f e a t u r e s i s g e n e r a l l y used, making the a l g o r i t h m l e s s s e n s i t i v e to departures from the data requirements. I n s e l e c t i n g the sample data f o r the generation of the d i s c r i m i n a n t f u n c t i o n , i t was found that the best performance was achieved i f the sample was taken at an intermediate p o i n t i n the dose range used i n a given experiment. Table VI summarizes the r e s u l t s of two such experiments. The method performed q u i t e w e l l , producing a mean d e v i a t i o n from the manual c l a s s i f i c a t i o n s of AS = ±0.018 ( c a l c u l a t e d from r e s u l t s of 4 experiments). This d e v i a t i o n was somewhat higher than that seen f o r the colony screening method (AS = ±0.008), but was s t i l l comparable to e r r o r r a t e s f o r manual s c o r i n g . I f the f l a s k s from one dose p o i n t i n a t y p i c a l s u r v i v a l experiment i n v o l v i n g a t o t a l of 7 or 8 dose p o i n t s were used to generate the d i s c r i m i n a n t f u n c t i o n , 6-14% of the t o t a l number of parent c e l l l o c a t i o n s would have to be manually c l a s s i f i e d . This percentage i s comparable to t h a t r e q u i r e d by the colony screening procedure. However, these manual c l a s s i f i c a t i o n s would only have to be c a r r i e d out i n 1 or 2 s e l e c t e d f l a s k s , r a t h e r than i n every f l a s k . In a d d i t i o n , i f the number of dose p o i n t s per experiment were increased, the number of c o l o n i e s t h a t would have to be manually c l a s s i f i e d f o r the d i s c r i m i n a n t f u n c t i o n method would remain the same, whereas f o r the colony screening method i t would increase w i t h the number of f l a s k s i n v o l v e d . Table VI. Performance of the d i s c r i m i n a n t f u n c t i o n method of colony s c o r i n g Dose (Gy) Siz e of Data Set Manual S u r v i v a l QDF t S u r v i v a l D e v i a t i o n Percent D e v i a t i o n Experiment #2 0 404 0 866 0 851 -0 015 -1 7 0.5 457 0 796 0 816 0 020 2 4 1.2* 500 0 658 0 656 -0 002 -0 3 2.0 459 0 490 0 508 0 018 3 3 Experiment #2 0 309 0 896 0 880 -0 016 -1 8 0.4 299 0 773 0 799 0 026 3 4 0.8 275 0 709 0 745 0 036 4 9 1.2* 293 0 580 0 577 -0 003 -0 6 1.6 307 0 550 0 570 0 020 3 4 2.0 296 0 446 0 459 0 013 2 9 2.4 328 0 352 0 358 0 006 1 7 t s u r v i v a l c a l c u l a t e d u s i n g a quadratic d i s c r i m i n a n t f u n c t i o n (QDF) to c l a s s i f y s u r v i v o r s and non-survivors data p o i n t used to generate d i s c r i m i n a n t f u n c t i o n 154 Despite the f a c t that the number of c o l o n i e s that had to be c l a s s i f i e d manually d i d not increase as the s i z e of the experiment increased, the d i s c r i m i n a n t f u n c t i o n method s t i l l r e t a i n e d the disadvantage th a t colony scans had to be c a r r i e d out on two days, i n s t e a d of j u s t one. However, day 2 colony i n f o r m a t i o n could prove u s e f u l f o r purposes other than those d i r e c t l y r e l a t e d to the assessment of c e l l s u r v i v a l . For i n s t a n c e , s i n c e e s s e n t i a l l y a l l c e l l s i n an i r r a d i a t e d p o p u l a t i o n undergo at l e a s t one d i v i s i o n (Brosing, 1983; Grote et a l , 1981a; E l k i n d et a l , 1963), day 2 colony scans could be used to f u r t h e r confirm c e l l c l a s s i f i c a t i o n s made by the r e c o g n i t i o n a l g o r i t h m on the day of p l a t i n g . By p e r m i t t i n g a higher f a l s e p o s i t i v e e r r o r r a t e on the day of p l a t i n g , such a c o n f i r m a t i o n procedure might a l l o w the number of c e l l s m i s c l a s s i f i e d by the r e c o g n i t i o n a l g o r i t h m to be reduced. 6.6 Factors A f f e c t i n g Accuracy of P o p u l a t i o n Density Estimates 6.6.1 Focus Dependence of Population Density Measurements Several f a c t o r s a f f e c t e d the accuracy of p o p u l a t i o n d e n s i t y estimates made from colony f e a t u r e scores. Foremost among these was the focus s e t t i n g used when feature data were c o l l e c t e d . U n l i k e the s i t u a t i o n w i t h c e l l d e t e c t i o n and r e c o g n i t i o n (Sections 4.4 and 5.6), colony f e a t u r e data were c o l l e c t e d even i f images were g r o s s l y out of focus. There were almost always some regions of a colony that produced s i g n a l s exceeding e i t h e r the upper or lower thresholds on which feature c a l c u l a t i o n s were based, even i f the focus s e t t i n g deviated s u b s t a n t i a l l y from optimum l e v e l s . This would not present a problem i f c o r r e c t focus l e v e l s f o r p o p u l a t i o n d e n s i t y scans co u l d be e s t a b l i s h e d . However, because V79 c e l l s are f l a t t e n e d and grow i n c l o s e p r o x i m i t y i n a colony, an " i d e a l " focus l e v e l was d i f f i c u l t to define 155 f o r them. The problem was not as daunting f o r CHO c e l l s , s i n c e they remain d i s t i n c t and rounded even when they form c o l o n i e s , a l l o w i n g i d e a l focus l e v e l s to be e s t a b l i s h e d f o r colony scanning i n the same way they were found f o r s i n g l e c e l l s ( S e c t i o n 4.4.1). To determine the e f f e c t of focus s e t t i n g on V79 colony feature scores, a s e r i e s of feature measurements was made at d i f f e r e n t focus l e v e l s . As f o r the focus measurements made f o r c e l l d e t e c t i o n and r e c o g n i t i o n , data were c o l l e c t e d from only a s m a l l , r e l a t i v e l y f l a t r e g i o n of the t i s s u e c u l t u r e f l a s k . The " i d e a l " focus s e t t i n g c o u l d not be determined p r i o r to making the measurements, since no r e l a t i v e l y simple fe a t u r e of a colony l i n e scan could be used to c o n s i s t e n t l y i d e n t i f y t h i s s e t t i n g . Line scans taken at d i f f e r e n t l o c a t i o n s confined to even a s i n g l e colony d i d not always produce feature maxima at the same focus l e v e l . V a r i a t i o n s as l a r g e as ±30 pm were o f t e n seen. The "zero" ( i d e a l ) l e v e l f o r the focus data t h e r e f o r e had to be e s t a b l i s h e d from the e n t i r e focus dependence curve, as described below. The focus dependence f o r the " t o t a l area" colony f e a t u r e i s shown f o r c o l o n i e s of v a r i o u s ages i n Figure 49,a-e. These data were c o l l e c t e d on successive days from the same 3 f l a s k s . Data were pooled to form the p l o t s of the f i g u r e by matching the maxima and minima of the focus dependence curves f o r the i n d i v i d u a l f l a s k s . C l e a r l y , the focus dependence of the " t o t a l area" f e a t u r e f o l l o w s a s i m i l a r p a t t e r n to that seen f o r the s i g n a l peak height from a s i n g l e , rounded c e l l (Figure 22). The form of the focus dependence was s i m i l a r f o r a l l the colony features measured by SSCAN. However, the volume features (e.g. Figure 4 9 f ) , as w e l l as the number of peaks and the number of troughs, had a somewhat sharper focus dependence than the area features f o r c o l o n i e s of the same age. 156 (a) total area, day 1 -50 0 50 focus level (um) loo -100 -50 o focus level (pm) F i g u r e 49. Focus dependence o f some c o l o n y f e a t u r e s . (a) " t o t a l a r e a " , 1 - d a y - o l d c o l o n i e s , (b) " t o t a l a r e a " , 2 - d a y - o l d c o l o n i e s , ( c ) " t o t a l a r e a " , 3 - d a y - o l d c o l o n i e s , (d) " t o t a l a r e a " , 4 - d a y - o l d c o l o n i e s , (e) " t o t a l a r e a " , 5 - d a y - o l d c o l o n i e s , (a) " t o t a l volume", 4 - d a y - o l d c o l o n i e s . E r r o r b a r s r e p r e s e n t s t a n d a r d d e v i a t i o n s . 157 The sharpness of the focus dependence a l s o increased w i t h colony age. Along w i t h t h i s increased sharpness, the distance between the focus l e v e l t h a t produced maximum feature scores and the focus l e v e l t h a t produced minimum f e a t u r e scores decreased. In order to determine which, i f e i t h e r , of these two c h a r a c t e r i s t i c p o i n t s i n the focus dependence curve corresponded to the "same" focus l e v e l , data from a s i n g l e day (day 3) were d i v i d e d i n t o groups c o n t a i n i n g c o l o n i e s w i t h i n s p e c i f i c s i z e ranges (as measured by the " t o t a l area" f e a t u r e ) . A p l o t of the focus dependence of these groups revealed that the minimum feature scores a l l f e l l at approximately the same focus l e v e l , whereas the l o c a t i o n of maximum feature scores increased as the colony s i z e increased. A l l the focus dependence p l o t s of Figure 49 were the r e f o r e zeroed so that the f e a t u r e minimum f e l l at the same focus l e v e l . Consequently, feature maxima were l o c a t e d only approximately at the " i d e a l " ("0") focus l e v e l f o r some c o l o n i e s . From the above observations, i t was apparent t h a t the problem of p r o v i d i n g accurate focus adjustments f o r colony scanning was compounded by the f a c t t h a t c o l o n i e s of d i f f e r e n t s i z e d i d not produce maximum feature scores at the same focus l e v e l . M i n i m i z a t i o n of f e a t u r e v a l u e s , f o l l o w e d by the a p p l i c a t i o n of a 50-60 /xm negative o f f s e t , would l i k e l y provide more c o n s i s t e n t f o c u s i n g . Even then, i n f o r m a t i o n from a s i n g l e colony l i n e scan would not be s u f f i c i e n t f o r f o c u s i n g purposes. An autofocusing a l g o r i t h m t h a t r e q u i r e d i n f o r m a t i o n from an e n t i r e 256 X 256 p i x e l colony scan would, however, be very slow i f a b i n a r y search s i m i l a r to that used f o r s i n g l e c e l l s had to be employed. Thus, to date, a r e l i a b l e means of f o c u s i n g on V79 c o l o n i e s has not been implemented i n the C e l l Analyzer. Because no simple, o b j e c t i v e means f o r m a i n t a i n i n g c o r r e c t focus s e t t i n g s f o r colony scanning was a v a i l a b l e , manual f o c u s i n g was g e n e r a l l y used f o r colony measurements. An exception to t h i s was i n the c o l l e c t i o n 158 of the data f o r Figure 43, where focus l e v e l s were e s t a b l i s h e d f o r known f l a t regions of the f l a s k by maximizing the peak he i g h t s f o r round, i s o l a t e d s i n g l e c e l l s l o c a t e d among the c o l o n i e s . Otherwise, focus was f r e q u e n t l y monitored manually throughout the scanning of each f l a s k . This method provided s u f f i c i e n t l y c o n s i s t e n t colony f e a t u r e data to a l l o w the development of the colony c l a s s i f i c a t i o n schemes de s c r i b e d i n Sections 6.5.2 and 6.5.3. However, si n c e the SSCAN program scans each colony i n approximately the same time t h a t would be r e q u i r e d f o r an experimenter to r e v i s i t and manually c l a s s i f y i t , manual monitoring of the focus s e t t i n g under experimental c o n d i t i o n s would defeat the purpose of performing automated colony scans at a l l . The experimenter would spend the same amount of time (or more, i n the case of the d i s c r i m i n a n t f u n c t i o n method) monitoring the focus f o r the colony scans as i t would otherwise have taken to manually c l a s s i f y a l l the c o l o n i e s . An attempt was t h e r e f o r e made to develop a l e s s time-consuming, although n e c e s s a r i l y cruder, method of m a i n t a i n i n g proper focus l e v e l s during colony scans. The method f i n a l l y employed was s i m i l a r to the 3-band semi-automated method described i n S e c t i o n 4.4.4, w i t h the exception t h a t the 3 focus l e v e l s used to define the f l a s k shape were determined manually. The 3-point f o c u s i n g method was employed f o r the c o l l e c t i o n of colony data f o r s e v e r a l experiments. U n f o r t u n a t e l y , a n a l y s i s of these data revealed that n e i t h e r of the two methods developed f o r the automated c l a s s i f i c a t i o n of c o l o n i e s performed as w e l l as they had when focus was c o n s t a n t l y monitored during scanning. The diminished performance was suspected to be due to f o c u s i n g problems, not only because a cruder f o c u s i n g method had been used, but a l s o because examination of the feature data suggested that t h i s might be so. In p a r t i c u l a r , values of the r a t i o 159 of volume above to area above (va/aa), which i s a measure of the mean s i g n a l i n t e n s i t y above background and i s th e r e f o r e r e l a t i v e l y independent of colony s i z e (Figure 50), were examined f o r f l a s k s i n v o l v e d i n s e v e r a l experiments. This a n a l y s i s i n d i c a t e d that focus s e t t i n g s were not very c o n s i s t e n t , both between f l a s k s and even w i t h i n i n d i v i d u a l f l a s k s . Indeed, f o r the d i s c r i m i n a n t f u n c t i o n method of colony c l a s s i f i c a t i o n , a r e l a t i v e l y good c o r r e l a t i o n between i n c o n s i s t e n t va/aa values and l a r g e e r r o r s i n s u r v i v a l estimates was found. E r r o r s i n the s u r v i v a l estimates made by the colony screening procedure were not as la r g e as those seen u s i n g the d i s c r i m i n a n t f u n c t i o n method, which r e l i e d on two sets of colony measurements, both of which were subje c t to i n c o r r e c t or i n c o n s i s t e n t f ocusing. Indeed, i t was found t h a t , by decreasing the lower l i m i t of the screening i n t e r v a l from 1000 to 500 ( " t o t a l area" f e a t u r e ) , reasonably accurate estimates of s u r v i v a l c o u l d be obtained even from the p o o r l y focused data. This d i d , however, increase the number of c o l o n i e s that had to be manually scored from 10% to approximately 30% of the data set. Nevertheless, the experimenter would s t i l l be spared a considerable amount of time and e f f o r t on manual c l a s s i f i c a t i o n . 6.6.2 Other Factors A f f e c t i n g Accuracy of Population Density Measurements Factors other than focus s e t t i n g a l s o a f f e c t e d the accuracy of po p u l a t i o n d e n s i t y estimates. Among these was the presence of d e b r i s . Large pieces of f i b r o u s debris could, f o r example, almost double scores obtained f o r those features measured below background l i g h t l e v e l s ( i . e . "number of troughs", "area below", and "volume below"). The " t o t a l area" and " t o t a l volume" features were a l s o a f f e c t e d , but to a l e s s e r extent. The presence of f i b r o u s debris was hard to c o n t r o l , as i t s source has 10 A ta>5000 U 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r — i 1 -100 -50 0 50 100 focus level (pm) F i g u r e 50. The r a t i o , "volume a b o v e " / " a r e a above" ( v a / a a ) , as a f u n c t i o n o f f o c u s s e t t i n g . Focus dependence i s shown f o r c o l o n i e s o f v a r i o u s s i z e , where c o l o n y s i z e has been d e f i n e d a c c o r d i n g t o the " t o t a l a r e a " ( t a ) f e a t u r e v a l u e a t i d e a l f o c u s ("0" f o c u s l e v e l on g r a p h ) . Each c u r v e r e p r e s e n t s t h e mean v a l u e f o r s e v e r a l c o l o n i e s . 161 remained unknown. I t was, however, easy to spot under the microscope, a l l o w i n g appropriate a c t i o n ( i . e . manual s c o r i n g of c o l o n i e s ) to be taken. Another f a c t o r that c o u l d a f f e c t colony f e a t u r e values f o r V79 c e l l s was the temperature of the f l a s k . As already i n d i c a t e d i n Figure 40, c e l l s i n V79 c o l o n i e s begin to round up i f l e f t at room temperature. This could have a l a r g e e f f e c t on colony f e a t u r e measurements, as shown f o r day 4 c o l o n i e s i n Figure 51. Both data sets i n t h i s f i g u r e were c o l l e c t e d from the same group of c o l o n i e s . No s i g n i f i c a n t c e l l growth would have occurred during the 5 hours at room temperature, since mammalian c e l l s do not p r o l i f e r a t e at sub-optimal temperatures. Nevertheless, f e a t u r e values at optimal focus l e v e l s changed by as much as a f a c t o r of 3 f o r some colony f e a t u r e s during t h i s p e r i o d . Thus, colony measurements r e q u i r i n g more than a few minutes had to be c a r r i e d out on a microscope stage equipped w i t h an incubator. This was, however, u s u a l l y not necessary i n s u r v i v a l experiments, s i n c e each f l a s k had to be out of the incubator f o r only a 5-8 minute scanning p e r i o d . 162 (a) volume above o 20-22 °C • 37 ° C (c) total area O X o 20 -22 ° C • 37 ° C -100 -50 0 focus level (um) Figure 51. E f f e c t of a 5-hour p e r i o d at room temperature on colony feature scores. (a) "volume above", (b) "volume below", (c) " t o t a l area". The increase i n the feat u r e scores a f t e r time spent at room temperature i s due to rounding up of the c e l l s ( c f . Figure 40). 163 7. APPLICATION TO LOW DOSE RADIATION STUDIES: MEASUREMENT OF THE RELATIVE  BIOLOGICAL EFFECTIVENESS OF DIFFERENT LOW LET RADIATIONS 7.1 M o t i v a t i o n f o r Measuring the RBE of Low LET R a d i a t i o n s at Low Doses 7.1.1 Influence of LET on B i o l o g i c a l E f f e c t i v e n e s s As discussed i n S e c t i o n 1.2.2, the p a t t e r n of energy d e p o s i t i o n produced by i o n i z i n g r a d i a t i o n s i n f l u e n c e s t h e i r r e l a t i v e e f f e c t i v e n e s s . Numerous s t u d i e s w i t h d i f f e r e n t b i o l o g i c a l systems have shown th a t r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s (RBE) increases w i t h i n c r e a s i n g l i n e a r energy t r a n s f e r (LET) (Skarsgard et a l , 1967; Barendsen, 1968). Since i t s i n t r o d u c t i o n by Z i r k l e et a l (1952), the concept of LET has been widely used i n r a d i o b i o l o g y and r a d i a t i o n p r o t e c t i o n to s p e c i f y r a d i a t i o n q u a l i t y . The c h a r a c t e r i s t i c shape of an RBE versus LET curve f o r mammalian c e l l s u r v i v a l i s i l l u s t r a t e d i n Figure 52. RBE increases w i t h i n c r e a s i n g LET, f i r s t g r a d u a l l y , then r i s i n g sharply to a maximum at approximately 100 keV//im (Skarsgard et a l , 1967; Barendsen, 1968). Beyond t h i s maximum, RBE decreases. The decrease i s g e n e r a l l y a t t r i b u t e d to " o v e r k i l l " , which i s thought to occur when i o n i z i n g events are so c l o s e l y spaced that more energy than i s necessary f o r c e l l k i l l i n g i s deposited (Barendsen, 1968). The h i g h i o n i z a t i o n d e n s i t y at h i g h LET can a l s o increase the p r o b a b i l i t y of recombination events between r e a c t i v e species ( e l e c t r o n s , i o n s , and free r a d i c a l s ) , thereby causing them to be e l i m i n a t e d before they have the opportunity to cause b i o l o g i c a l damage ( K r a f t and Kraft-Weyrather, 1987). While the shape of the RBE versus LET curve i s s i m i l a r f o r a v a r i e t y of b i o l o g i c a l systems, absolute RBE values can be a f f e c t e d by v a r i o u s f a c t o r s . In a d d i t i o n to the b i o l o g i c a l system or endpoint used, these f a c t o r s i n c l u d e dose r a t e , dose f r a c t i o n a t i o n , and the l e v e l of e f f e c t at 164 10° 101 10 2 10 3 LET (keV/pm) F i g u r e 52. C h a r a c t e r i s t i c shape o f the RBE v e r s u s LET c u r v e f o r mammalian c e l l s . The most e f f i c i e n t k i l l i n g o c c u r s a t LET's o f a p p r o x i m a t e l y 100 keV//im. 165 which RBE i s measured. In general, RBE's f o r high LET r e l a t i v e to low LET r a d i a t i o n s decrease as the s u r v i v a l l e v e l at which they are measured decreases (Barendsen, 1968). I f the dose r a t e i s reduced or i f f r a c t i o n a t e d doses are a p p l i e d , the RBE of high LET r a d i a t i o n s increases f o r a given t o t a l dose (Barendsen, 1968; Z e i t z et a l , 1977). The use of LET as a means of s p e c i f y i n g r a d i a t i o n q u a l i t y has met w i t h some c r i t i c i s m . In a d d i t i o n to the f a c t t h a t the d e r i v a t i o n of a s i n g l e LET value to c h a r a c t e r i z e a p a r t i c u l a r r a d i a t i o n modality i n v o l v e s the l i n e a r averaging of q u a n t i t i e s that may not be d i r e c t l y p r o p o r t i o n a l to b i o l o g i c a l e f f e c t , t h i s c r i t i c i s m i s a l s o d i r e c t e d at the f a c t that LET cannot always be d i r e c t l y r e l a t e d to l o c a l energy d e p o s i t i o n i n a volume of m a t e r i a l . A simple r e l a t i o n s h i p does not e x i s t because of the e f f e c t s of t r a c k curvature, f i n i t e t r a c k l e n g t h , changes i n LET during p a r t i c l e t r a v e r s a l , escape of energy v i a d e l t a rays (secondary e l e c t r o n t r a c k s a r i s i n g from p a r t i c u l a r l y v i o l e n t c o l l i s i o n s i n the primary t r a c k ) , a d d i t i o n of energy from d e l t a rays generated by d i s t a n t t r a c k s , and s t a t i s t i c a l f l u c t u a t i o n s i n the energy l o s s events along the t r a c k (termed energy l o s s s t r a g g l i n g ) (Barendsen, 1968; K e l l e r e r and Chmelevsky, 1975; ICRU, 1986; Goodhead, 1987). Indeed, i t has been suggested that LET values can adequately p r e d i c t energy d e p o s i t i o n i n a reference volume only f o r heavie r charged p a r t i c l e s such as heavy ions and, w i t h some l i m i t a t i o n s , protons ( K e l l e r e r and Chmelevsky, 1975; R o s s i , 1986). This excludes a l l r a d i a t i o n s which cause damage through the d e p o s i t i o n of energy by e l e c t r o n s , and may provide some ex p l a n a t i o n f o r the f a c t that RBE's observed f o r these low LET r a d i a t i o n s increase only moderately even f o r r e l a t i v e l y l a r g e increases i n LET (ICRU, 1986). Other observations that i l l u s t r a t e the l i m i t a t i o n s of the LET concept i n p r e d i c t i n g RBE include those which show that the RBE-LET curve depends 166 on the type of r a d i a t i o n used. For p a r t i c u l a t e r a d i a t i o n s , f o r example, the maximum RBE appears to s h i f t towards higher LET values w i t h i n c r e a s i n g p a r t i c l e mass ( K r a f t and Kraft-Weyrather, 1987; B e l l i et a l , 1989; F o l k a r d et a l , 1989). One e x p l a n a t i o n f o r such observations i s t h a t , because l i g h t e r p a r t i c l e s have lower v e l o c i t i e s than h e a v i e r p a r t i c l e s at the same LET, the energy of the secondary e l e c t r o n s (and, t h e r e f o r e , the t r a c k s t r u c t u r e ) i s d i f f e r e n t ( F o l k a r d et a l , 1989). 7.1.2 Microdosimetry and R a d i a t i o n Q u a l i t y Because of the apparent l i m i t a t i o n s of the LET concept, i t has been suggested that microdosimetric parameters may provide a b e t t e r d e s c r i p t i o n of r a d i a t i o n q u a l i t y ( R o s s i , 1977, 1978; ICRU, 1986). Such parameters have even been in c o r p o r a t e d i n t o a model of r a d i a t i o n a c t i o n t h a t attempts to provide a l i n k between r a d i a t i o n physics and observed r a d i o b i o l o g i c a l phenomena. This model, developed by K e l l e r e r and R o s s i (1972), i s known as the Theory of Dual R a d i a t i o n A c t i o n , or TDRA. The TDRA has succeeded i n p r o v i d i n g explanations f o r some observed changes i n RBE w i t h r a d i a t i o n dose, most notably the l a r g e increase i n the RBE of neutrons r e l a t i v e to X-rays w i t h decreasing dose. This e f f e c t has been measured f o r a v a r i e t y of b i o l o g i c a l systems and endpoints ( K e l l e r e r and R o s s i , 1971, 1972 c o n t a i n a summary of the data). The TDRA p r e d i c t s a l i n e a r - q u a d r a t i c response f o r c e l l s u r v i v a l as a f u n c t i o n of r a d i a t i o n dose. The l i n e a r - q u a d r a t i c r e l a t i o n , which a l s o appears i n other models of r a d i a t i o n a c t i o n (e.g. Chadwick and Leenhouts, 1973), i s commonly expressed as: In S = -aD-/3D2 (7.1) 167 where D i s the dose and S i s the s u r v i v i n g f r a c t i o n . The l i n e a r (a) term i s i n t e r p r e t e d as r e p r e s e n t i n g l e t h a l l e s i o n s that a r i s e through the a c t i o n of a s i n g l e p a r t i c l e t r a c k , while the quadratic (/3) term i s i n t e r p r e t e d as a r i s i n g from the i n t e r a c t i o n of two n o n - l e t h a l " s u b l e s i o n s " (created by d i f f e r e n t p a r t i c l e t r a c k s ) to form a l e t h a l l e s i o n . In the TDRA, the c o e f f i c i e n t s a and j3 are r e l a t e d to a microdosimetric q u a n t i t y known as the mean s p e c i f i c energy, f, deposited i n a " s e n s i t i v e s i t e " w i t h i n the c e l l (see Appendix, S e c t i o n 10.2 f o r a more thorough o u t l i n e of the u n d e r l y i n g p r i n c i p l e s and assumptions of the TDRA). In terms of a and /3: a = k c f3 = k f = a/B (7.2) (7.3) (7.4) where k i s a p r o p o r t i o n a l i t y constant r e l a t e d to the e f f i c i e n c y of l e s i o n production. Under c e r t a i n c o n d i t i o n s (e.g. uniform d i s t r i b u t i o n of dose i n the s e n s i t i v e s i t e , i o n i z a t i o n d e n s i t y below l e v e l a t which " o v e r k i l l " e f f e c t o c c u r s ) , k i s considered to be independent of r a d i a t i o n q u a l i t y ( K e l l e r e r and R o s s i , 1972). These c o n d i t i o n s apply to most r a d i a t i o n s of low to moderate LET. The TDRA assumes that only those sublesions that a r i s e w i t h i n a s e n s i t i v e s i t e of some given s i z e can i n t e r a c t to form l e t h a l l e s i o n s . The diameter, d, of the s e n s i t i v e s i t e defines the maximum dis t a n c e t h a t can separate two i n t e r a c t i n g s u b l e s i o n s . I t can be r e l a t e d to the s p e c i f i c energy, f ( i n Gy), by: c - 0.204 yD / d 2 (7.5) where i s the dose-averaged l i n e a l energy ( i n keV/pm) f o r a given 168 r a d i a t i o n modality, and d i s measured i n /xm ( K e l l e r e r and R o s s i , 1972). When a l i n e a r - q u a d r a t i c model i s used to describe the r a d i a t i o n ' dose-response, the RBE of d i f f e r e n t r a d i a t i o n m o d a l i t i e s at a given l e v e l of e f f e c t ( i . e . at a given s u r v i v a l l e v e l ) can be c a l c u l a t e d from a and /3: Di -a r + [a. 2 + 4/3rD (a + /3D)] 1 / 2 RBE - • (7.6) D 2pD where D and D are the doses of the reference and t e s t r a d i a t i o n s , r r e s p e c t i v e l y , a t the s e l e c t e d s u r v i v a l l e v e l . S i m i l a r l y , c*r and /3 are the c o e f f i c i e n t s f o r the reference r a d i a t i o n , and a and /3 are the c o e f f i c i e n t s f o r the t e s t r a d i a t i o n . Low and high dose l i m i t s can be deduced from Equation 7.6: li r a RBE = a/a. and lim RBE = (/3//3r)1/2. (7.7) D -* 0 D •* oo Thus, i f mammalian c e l l s u r v i v a l curves are c o r r e c t l y represented by a l i n e a r - q u a d r a t i c equation, the RBE i n the low and high dose ranges i s given by r a t i o s of the l i n e a r or quadratic c o e f f i c i e n t s . I n t e r p r e t e d i n terms of the TDRA, the RBE at low doses approaches the value obtained by c a l c u l a t i n g the r a t i o of f's f o r the two m o d a l i t i e s i n question i f k i s independent of r a d i a t i o n q u a l i t y . This theory t h e r e f o r e allows RBE to be d i r e c t l y r e l a t e d to a measurable microdosimetric q u a n t i t y . Furthermore, the proposed constancy of k leads to the p r e d i c t i o n that the RBE at high doses approaches a value of 1.0. 169 7.1.3 RBE at Low Doses Because i t u t i l i z e s measurable microdosimetric q u a n t i t i e s , the TDRA allows c e r t a i n q u a n t i t a t i v e p r e d i c t i o n s to be made about the response of b i o l o g i c a l systems to d i f f e r e n t r a d i a t i o n m o d a l i t i e s . As already mentioned i n S e c t i o n 7.1.2, t h i s was demonstrated by K e l l e r e r and R o s s i (1972) usi n g data measuring the dependence of RBE on dose f o r neutrons versus X-rays. Large increases i n RBE w i t h decreasing dose co u l d be ex p l a i n e d f o r these data by the l a r g e C r a t i o f o r the two m o d a l i t i e s . A trend towards an RBE of 1 at h i g h doses (up to 10 Gy) was a l s o seen. S i m i l a r agreement w i t h the theory was l a t e r obtained u s i n g 250 kVp X-rays and neutrons of d i f f e r e n t energies on V79 c e l l s i n t i s s u e c u l t u r e ( H a l l et a l , 1975; K e l l e r e r et a l , 1976) . A f u r t h e r p o i n t that has been used to support the theory i s that s i t e diameters (d) c a l c u l a t e d from ex p e r i m e n t a l l y d e r i v e d C values f a l l w i t h i n a narrow range (1-3 pm) f o r a la r g e number of b i o l o g i c a l systems, suggesting t h a t there may be a common s e n s i t i v e s i t e amongst them ( K e l l e r e r and R o s s i , 1972). The p r e d i c t i o n s of the TDRA have r a m i f i c a t i o n s not only f o r the dose dependence of RBE f o r high LET r a d i a t i o n s such as neutrons, but a l s o f o r low LET r a d i a t i o n s that are commonly encountered both environmentally and c l i n i c a l l y . "Low LET" r a d i a t i o n s i n c l u d e X- and 7-rays, as w e l l as d i r e c t l y i o n i z i n g p a r t i c l e s w i t h LET l e s s than or equal to 3.5 keV/pm (ICRP, 1955). RBE's f o r c e l l s u r v i v a l a f t e r exposure to these r a d i a t i o n s have been measured to be between 1.0 and 1.5 (see Table V I I ) . Because conventional s u r v i v a l assays were used to gather the data, however, these RBE values were determined at r e l a t i v e l y h i g h doses and l e v e l s of e f f e c t , where the quadratic component of a l i n e a r - q u a d r a t i c dose response would predominate. RBE's f o r c e l l s u r v i v a l curves are t y p i c a l l y c a l c u l a t e d at 170 Table V I I . B i o l o g i c a l e f f e c t i v e n e s s ( r e l a t i v e to 6 0 C o 7 - r a y s t ) of s e l e c t e d low LET r a d i a t i o n s : d i r e c t observations and p r e d i c t i o n s based on the Theory of Dual R a d i a t i o n A c t i o n R a d i a t i o n modality Observed RBE's from conventional h i g h dose s u r v i v a l experiments RBE's measured at low doses f o r c e r t a i n s e n s i t i v e b i o l o g i c a l systems Low dose RBE p r e d i c t e d from mic r o d o s i m e t r i c measurements 30-60 kV p X-rays 1.25 - 1.55 a -7 b t -3.5 c 200-300 kV„ X-rays 1.1 d 2 - 3 e 2.5 - 3.0 c 3-18 MeV e l e c t r o n s -0.9 f t 0.2 - 0.7 St 0.6 - 0.8 h a from Z e i t z et a l (1977) (HeLa c e l l s , maximum dose 7 Gy) and Hoshi et a l (1988) (V79 c e l l s , maximum dose 12 Gy), w i t h 2-4 Gy dose i n t e r v a l s . RBE's were estimated from data at 10% s u r v i v a l or lower. k i n d u c t i o n of d i c e n t r i c s i n human lymphocytes by 30 kV X-rays ( V i r s i k et a l , 1977). c from E l l e t and Braby (1972) and Bond et a l (1978). ^ from Malone et a l (1974) f o r HeLa and CHO c e l l s u r v i v a l (maximum dose 10 Gy, -2 Gy dose i n t e r v a l s , RBE determined at 10% s u r v i v a l or l e s s ) . Supported by RBE's of 1.1-1.2 f o r other systems and endpoints ( S i n c l a i r , 1962; S i n c l a i r and Kohn, 1964; W i l l i a m s and Hendry, 1978). e i n d u c t i o n of d i c e n t r i c s i n human lymphocytes by 250 kV X-rays (maximum dose 5 Gy, 0.5-1.0 Gy dose i n t e r v a l s ) (LLoyd et a l , 19/5). RBE was determined from a - r a t i o s . Supported by data from Underbrink et a l (1976) f o r mutation of Tradescantia stamen h a i r s (an a - r a t i o of 2.1 was found f o r 250 kVp X-rays r e l a t i v e to 1 3 7 C s 7-rays; maximum dose 1 Gy). f based on data from Amols et a l (1986) f o r 10% s u r v i v a l of DLD-1 c e l l s (human co l o n carcinoma) exposed to 18 MeV e l e c t r o n s . Supported by data from v a r i o u s other b i o l o g i c a l systems and endpoints (Williams and Hendry, 1978; S i n c l a i r and Kohn, 1964). S based on e x t r a p o l a t i o n s from h i g h dose c e l l s u r v i v a l data (Amols et a l , 1986) and on a - r a t i o f o r i n d u c t i o n of d i c e n t r i c s i n human lymphocytes by 3 MeV e l e c t r o n s (maximum dose 4 Gy) (Schmid et a l , 1974). h from Amols et a l (1986). t i n some in s t a n c e s , the reference r a d i a t i o n f o r the o r i g i n a l experiments represented i n t h i s t a b l e was 150-250 kVp X-rays. For these experiments, RBE's have been converted to the 6 0 C o 7-ray standard by m u l t i p l y i n g by 1.1 f o r h i g h dose RBE's and by 2.5 f o r low dose RBE's (on the b a s i s of the RBE values l i s t e d i n t h i s t a b l e f o r 200-300 kV X - r a y s ) . 171 10% s u r v i v a l , and e x t r a p o l a t i o n s to higher s u r v i v a l l e v e l s are not u s u a l l y made, presumably because of t h e i r dependence on the s u r v i v a l model used. I n c o n t r a s t to the measured RBE's f o r low LET r a d i a t i o n s , m i c r o d o s i m e t r i c determinations of f i n 1 /xm diameter spheres f o r low LET r a d i a t i o n s have i n d i c a t e d t h a t the low-dose RBE ( r e l a t i v e to 6 0 C o 7-rays) may be as h i g h as 2.5-3.0 f o r 200-300 kV p X-rays, and even higher (-3.5) f o r 30-60 kVp X-rays. Furthermore, measurements made at low doses w i t h c e r t a i n h i g h l y s e n s i t i v e b i o l o g i c a l systems have corresponded w e l l w i t h these p r e d i c t i o n s (the value f o r k i n Equations 7.2 and 7.3 a l s o appeared to be r e l a t i v e l y constant between r a d i a t i o n m o d a l i t i e s i n these experiments) . One of the systems used f o r such measurements was the f l o w e r i n g p l a n t , Tradescantia, the normally b l u e - c o l o u r e d stamen h a i r s of which are s u b j e c t to pink mutation events upon exposure to r a d i a t i o n . Other r e s u l t s that support the m i c r o d o s i m e t r i c a l l y - b a s e d p r e d i c t i o n s have been obtained i n s t u d i e s of chromosome a b e r r a t i o n s ( p a r t i c u l a r l y d i c e n t r i c s ) i n human lymphocytes. The r e s u l t s of the low dose measurements are i n c l u d e d i n Table V I I , as are the RBE p r e d i c t i o n s made from m i c r o d o s i m e t r i c a l measurements. The low-dose e f f e c t s seen i n Tradescantia and i n the formation of d i c e n t r i c s i n human lymphocytes could have i m p l i c a t i o n s f o r r a d i o p r o t e c t i o n and radiotherapy, e s p e c i a l l y i f they extend to other endpoints. Indeed, lymphocyte data c o l l e c t e d f o r a l a r g e range of LET'S have been used to d e r i v e an e m p i r i c a l r e l a t i o n s h i p between r a d i a t i o n q u a l i t y , Q, and l i n e a l energy, y, (Zaider and Brenner, 1985) that has been recommended f o r r a d i a t i o n p r o t e c t i o n purposes (ICRU, 1986). I f the low-dose e f f e c t s summarized i n Table V I I extend to c e l l s u r v i v a l , they may a f f e c t r a d i o t h e r a p e u t i c response, since the dose f r a c t i o n s commonly employed i n treatment regimens may be low enough that the l i n e a r component (and, 172 t h e r e f o r e , $*) dominates the response. Measurements of c e l l s u r v i v a l at s u f f i c i e n t l y low doses to i n v e s t i g a t e t h i s e f f e c t have been d i f f i c u l t to c a r r y out due to the l i m i t a t i o n s of conventional assays. Some attempts at performing e x t r a p o l a t i o n s w i t h h i g h dose and f r a c t i o n a t e d dose data u s i n g the l i n e a r - q u a d r a t i c equation have, however, suggested that the p r e d i c t i o n s of the TDRA co u l d extend to c e l l s u r v i v a l . These experiments (Amols e t a l , 1986) measured the RBE of megavoltage X-rays and e l e c t r o n s versus 250 kVp X-rays, u s i n g human tumour c e l l s u r v i v a l as the endpoint. The r e s u l t s ( a l s o summarized i n Table VII) suggested that the RBE's of megavoltage r a d i a t i o n s were co n s i d e r a b l y diminished at low doses, as p r e d i c t e d from microdosimetric measurements. E x t r a p o l a t i o n s , however, depend on the v a l i d i t y of the equations they are attempting to t e s t . Furthermore, i n a d d i t i o n to the experimental r e s u l t s supporting the p r e d i c t i o n s of the TDRA, data t h a t do not f i t c e r t a i n p r e d i c t i o n s of the theory have a l s o been c o l l e c t e d . Most of these r e s u l t s have been obtained u s i n g c e l l s u r v i v a l as the endpoint, but only i n the high-dose range. Included among the evidence c o n t r a d i c t i n g the TDRA are experiments w i t h u l t r a s o f t ( i . e . very low energy) X-rays, which produce extremely short ( « 1 um) p a r t i c l e t r a c k s of r e l a t i v e l y h i g h LET (20-30 keV//im). These low energy X-rays have f values comparable to those of 7-rays when measured i n 1 ixm diameter spheres, yet t h e i r b i o l o g i c a l e f f e c t i v e n e s s i s at l e a s t a f a c t o r of 2 higher (Cox et a l , 1977; Goodhead, 1977). Moreover, i n these same experiments, k-values were not found to be the same f o r the d i f f e r e n t m o d a l i t i e s . These d i s c r e p a n c i e s may, however, be a t t r i b u t e d a t l e a s t i n pa r t to the uneven dose d i s t r i b u t i o n expected i n a 1 /xm diameter s e n s i t i v e s i t e f o r such short p a r t i c l e t r a c k s . Experiments w i t h s p a t i a l l y c o r r e l a t e d ions ( B i r d , 1979), where p a i r s of p a r t i c l e t r a c k s separated by known distances were used to i r r a d i a t e 173 c e l l s , have a l s o revealed d i s c r e p a n c i e s w i t h the concept of a unique s e n s i t i v e s i t e w i t h dimensions on the order of micrometers. The c o r r e l a t e d i o n r e s u l t s have, however, been i n t e r p r e t e d according to a modifi e d form of the TDRA (see Appendix, S e c t i o n 10.2) that does not assume a s e n s i t i v e s i t e w i t h d i s t i n c t boundaries and a uniform p r o b a b i l i t y of s u b l e s i o n i n t e r a c t i o n w i t h i n those boundaries ( K e l l e r e r e t a l , 1979; K e l l e r e r and R o s s i , 1978). U n f o r t u n a t e l y , p r e d i c t i o n s of r a d i a t i o n response based s t r i c t l y on microdosimetric parameters cannot be made us i n g the mo d i f i e d theory. In order to determine whether the p r e d i c t i o n s of the TDRA regarding low LET r a d i a t i o n s apply to the doses and endpoints r e l e v a n t to radiotherapy, measurements would i d e a l l y have to made e x p l i c i t l y i n the r e l e v a n t dose range of 0-3 Gy. Since i t was p o s s i b l e to attempt such measurements usi n g the C e l l Analyzer, some of the low LET r a d i a t i o n m o d a l i t i e s commonly used i n c l i n i c a l a p p l i c a t i o n s were t e s t e d f o r t h e i r e f f e c t s on low-dose s u r v i v a l response. The s e l e c t e d m o d a l i t i e s were: 55 kVp X-rays, 250 kV p X-rays, 6 0 C o 7-rays, and 11 MeV e l e c t r o n s . Using these data, the TDRA cou l d be t e s t e d by: (1) determining d i r e c t l y the e f f e c t of dose on RBE i n the th e r a p e u t i c dose range, (2) checking whether the q u a d r a t i c term (/3, or k) of the f i t t i n g f u n c t i o n i s s i m i l a r f o r the d i f f e r e n t m o d a l i t i e s , and (3) assessi n g the v a l i d i t y of e x t r a p o l a t i o n s u s i n g a l i n e a r - q u a d r a t i c dose r e l a t i o n f i t t e d to high dose data. The l a t t e r assessment could be made by gathering h i g h dose data ( i . e . i n the 0-10 Gy range) c o n c u r r e n t l y w i t h data i n the low (0-3 Gy) dose range. 7.2 Materials and Methods 7.2.1 Radiation Apparatus and Dosimetry A l l experiments were performed w i t h c e l l s i n suspension i n s p e c i a l g l a s s i r r a d i a t i o n v e s s e l s (Parker et a l , 1969) r e f e r r e d to i n our l a b o r a t o r y as "ducks". During i r r a d i a t i o n , a duck would be seated i n a L u c i t e h o l d e r , which a l s o accommodated a thermometer and a 0.6 cm3 i o n i z a t i o n chamber (Figure 53). The i o n i z a t i o n chamber was used to monitor r a d i a t i o n exposure ("exposure" i s a measure of the charge l i b e r a t e d by i o n i z i n g r a d i a t i o n per u n i t mass of a i r , and i s p r o p o r t i o n a l to r a d i a t i o n dose under appropriate c o n d i t i o n s ) . An overhead magnetic s t i r r e r was attached to the L u c i t e h o lder so that c e l l s c o u l d be maintained i n suspension by continuous s t i r r i n g . I r r a d i a t i o n s were performed w i t h radiotherapy u n i t s c u r r e n t l y i n use f o r cancer treatment at the Cancer C o n t r o l Agency of B r i t i s h Columbia. The sources and p r o p e r t i e s of the r a d i a t i o n s used are summarized i n Table V I I I . Included i n t h i s t a b l e are estimates of dose u n i f o r m i t y across the depth of the c e l l suspension i n the duck. Of the four s e l e c t e d m o d a l i t i e s , 55 kVp X-rays, due to t h e i r low energy, are absorbed at the h i g h e s t r a t e i n t h e i r passage through v a r i o u s m a t e r i a l s . This caused a s i g n i f i c a n t s h i f t of the beam energy spectrum towards, higher energies as the beam passed through the L u c i t e holder and bottom surface of the g l a s s duck (shown i n Figure 54). The e f f e c t of f i l t r a t i o n by the duck and the L u c i t e h o lder on the 250 kVp X-ray beam i s a l s o shown f o r comparison. In c o n t r a s t to the e f f e c t s of a t t e n u a t i o n on the 55 kVp X-ray beam, the a t t e n u a t i o n of the 250 kVp X-ray beam was r e l a t i v e l y uniform across the energy spectrum, r e s u l t i n g i n only a small change i n the e f f e c t i v e beam energy. 1 7 5 Irradiation Vessel Thermometer 0.6 cc Ion Chamber Holder-TTTTTT1 Radiation F i g u r e 53. Diagram o f the e x p e r i m e n t a l s e t - u p u s e d f o r m e a s u r i n g t he RBE o f low LET r a d i a t i o n s . Sample a l i q u o t s o f the c e l l s u s p e n s i o n c o u l d be removed from t he i r r a d i a t i o n v e s s e l t h r o u g h t he s t o p p e r e d arm ( l e f t - h a n d s i d e o f f i g u r e ) . The 0.6 cm 3 i o n chamber was c o n n e c t e d t o a V i c t o r e e n e l e c t r o m e t e r f o r r a d i a t i o n e x p o s u r e measurements. The s e t - u p f o r beam d o s i m e t r y was i d e n t i c a l , w i t h t he e x c e p t i o n t h a t t he i r r a d i a t i o n v e s s e l c o n t a i n e d F r i c k e ( F r i c k e and Morse, 1929) s o l u t i o n i n p l a c e o f the c e l l s u s p e n s i o n . 176 Table V I I I . C h a r a c t e r i s t i c s of r a d i a t i o n sources used f o r RBE experiments M o d a l i t y 55 kV p X-rays 250 kV p X-rays 6 0 C o 7-rays 11 MeV e l e c t r o n s Source P h i l i p s RT 100 P h i l i p s RT 250 AECL Eldorado Super G Siemens Mevatron 12 l i n e a r a c c e l e r a t o r F i l t r a t i o n 1 mm Be + 0.78 mm A l 0.35 mm Cu + 0.40 mm Sn - -E f f e c t i v e Beam Energy a 20 keV (25 keV i n -s i d e duck) 122 keV 1.25 MeV 11 MeV Estimated change i n dose across depth of medium i n duck a , b -22% -5% -4% +3% LET,,, c 5.5 - 6.0 keV/pm 1.7 keV/um -0.24 keV/fxm 0.2 keV/um y n (averaged over 1 um) d not a v a i l a b l e 3.8 keV/>m 1.6 keV/fim 1.8 keV/um f f o r 1 /xm diameter s i t e s e -1.22 Gy -0.70 Gy -0.35 Gy not a v a i l a b l e a data provided by the Physics D i v i s i o n , Cancer C o n t r o l Agency of B r i t i s h Columbia b c a l c u l a t e d from percent depth-dose curves f o r 10 X 10 cm f i e l d s i z e . Note: 11 MeV e l e c t r o n dose increases across depth of medium i n duck because f u l l dose-buildup has not occurred at the p o i n t where the beam enters the duck. c from ICRU 16 (1970a) d from Zaider and Brenner (1985) e from Bond et a l (1978) OQ a © o O > (a) 55 kVp •unattenuated \ —attenuated 5 10 15 20 25 30 35 40 45 50 55 photon energy (keV) (b) 250 kVp •unattenuated * mmmM # AT • AT * x t # #/ t# »/ *# *# • / *# mm »# *# % —attenuated * V. V. 50 100 150 200 photon energy (keV) 250 F i g u r e 54. C a l c u l a t e d e nergy s p e c t r a f o r the 55 k V p and 250 k V p X - r a y beams. F o r b o t h c a s e s , s p e c t r a a r e shown f o r t h e u n a t t e n u a t e d beam and f o r the beam a f t e r i t has been a t t e n u a t e d by the L u c i t e h o l d e r and the bottom s u r f a c e o f the g l a s s i r r a d i a t i o n v e s s e l . P r o v i d e d by t h e P h y s i c s D i v i s i o n , C ancer C o n t r o l Agency o f B r i t i s h C o l u mbia. 178 The 55 kVp X-ray beam was a l s o attenuated s i g n i f i c a n t l y as i t passed through the c e l l suspension, as i n d i c a t e d i n Table V I I I . Despite the a t t e n u a t i o n of the beam, however, i t was presumed t h a t a l l c e l l s i n the s t i r r e d suspension would r e c e i v e the same dose during the course of i r r a d i a t i o n periods of reasonable d u r a t i o n . Because the c e l l s were to be i r r a d i a t e d i n suspension, dosimetry f o r the d i f f e r e n t r a d i a t i o n sources was performed u s i n g an aqueous chemical dosimeter: the fe r r o u s sulphate, or F r i c k e , dosimeter ( F r i c k e and Morse, 1929). This method of dosimetry can be used to a c c u r a t e l y estimate absorbed dose i n l i q u i d s (ICRU, 1969, 1970b, 1984). I t operates on the p r i n c i p l e t h a t , under c o n t r o l l e d c o n d i t i o n s , the o x i d a t i o n of f e r r o u s ions upon exposure to r a d i a t i o n i s d i r e c t l y p r o p o r t i o n a l to absorbed dose. F e r r i c i o n c o n c e n t r a t i o n i n the exposed s o l u t i o n was measured by spectrophotometric a n a l y s i s and converted to a dose estimate as described i n ICRU Reports 14 (1969), 17 (1970b), and 35 (1984). The procedure f o l l o w e d f o r beam dosimetry was as f o l l o w s : (1) With set-up as i n Figure 53, expose the F r i c k e s o l u t i o n to a s e l e c t e d r a d i a t i o n modality (a t o t a l dose of approximately 30 Gy was r e q u i r e d to o b t a i n a c c u r a t e l y measurable changes i n the f e r r i c i o n c o n c e n t r a t i o n of the s o l u t i o n ) . Monitor the t o t a l exposure measured by the i o n i z a t i o n chamber during the i r r a d i a t i o n . Record a l s o the t o t a l i r r a d i a t i o n time. (2) Measure the f e r r i c i o n c o n c e n t r a t i o n i n the F r i c k e s o l u t i o n , and from t h i s value determine the dose to the s o l u t i o n . (3) C a l c u l a t e the c o r r e l a t i o n f a c t o r between the dose to the s o l u t i o n i n the duck and the exposure measured by the i o n i z a t i o n chamber during the same i r r a d i a t i o n p e r i o d ( t h i s c o r r e l a t i o n f a c t o r w i l l be r e f e r r e d to as the exposure-to-dose conversion f a c t o r ) . 179 (4) C a l c u l a t e the dose r a t e from the measured dose to the s o l u t i o n i n the duck and the t o t a l i r r a d i a t i o n time. The dose ra t e s and the exposure-to-dose conversion f a c t o r s f o r the d i f f e r e n t r a d i a t i o n sources are l i s t e d i n Table IX. The c a l c u l a t e d dose r a t e s were used to determine the exposure times necessary to administer a given dose to the c e l l suspension during an experiment. The exact dose administered during any given i n t e r v a l was, however, c a l c u l a t e d by a p p l y i n g the exposure-to-dose conversion f a c t o r to exposure readings taken w i t h the i o n i z a t i o n chamber during that i n t e r v a l . These dose estimates were more accurate than those c a l c u l a t e d from the dose r a t e v a l u e s , s i n c e the a c t u a l dose r a t e f o r some of the radiotherapy u n i t s was only approximately constant w i t h time. The o v e r a l l e r r o r i n the doses determined using the exposure-to-dose conversion f a c t o r was estimated to be +2%, based on s e v e r a l r e p e t i t i o n s of the F r i c k e dosimetry measurements f o r each r a d i a t i o n modality. 7.2.2 C e l l P r e p a r a t i o n and I r r a d i a t i o n Procedure Because the low dose s u r v i v a l experiments were to be c a r r i e d out u s i n g the C e l l Analyzer, the amount of data that c o u l d be gathered on a s i n g l e day was l i m i t e d by the number of f l a s k s that c o u l d be comfortably scanned i n an 8-10 hour time p e r i o d , as p r e v i o u s l y discussed i n S e c t i o n 6.5.2. This c o n s t r a i n e d the number of s u r v i v a l curves that could be generated i n a s i n g l e experiment to two. A l l four r a d i a t i o n m o d a l i t i e s c o u l d t h e r e f o r e not be t e s t e d i n a s i n g l e experiment. Consequently, i n order to p r o p e r l y gauge the day-to-day v a r i a b i l i t y to which c e l l p o pulations are s u b j e c t , one of the four m o d a l i t i e s was chosen as a reference to which the other m o d a l i t i e s could be compared. A s u r v i v a l curve was then generated f o r t h i s reference modality i n each experiment, Table I X . Dose c a l c u l a t i o n f a c t o r s f o r the r a d i a t i o n sources used f o r RBE experiments Modality Exposure-to-dose conversion factort Approximate dose rate in duck 55 kV p X-rays 0.378 Gy/R 1.1 Gy/minute 250 kVp X-rays 0.833 Gy/R 1.8 Gy/minute 6 0 C o 7-rays 0.873 Gy/R 3.4 Gy/minute 11 MeV e l e c t r o n s 0.874 Gy/R 9 Gy/minute the conversion f a c t o r determines the r e l a t i o n s h i p between exposure measured w i t h the i o n i z a t i o n chamber to r a d i a t i o n dose i n the c e l l suspension (as measured by F r i c k e dosimetry). Exposure i s a measure the a b i l i t y of r a d i a t i o n to i o n i z e a i r , and i s given i n u n i t s of Roentgens (R) , where 1 R = 2.58 x 10"'' coulombs/kg of a i r . 181 along w i t h a curve f o r one of the remaining 3 " t e s t " m o d a l i t i e s . The s e l e c t e d reference r a d i a t i o n was 6 0 C o 7-rays, p r i m a r i l y because t h i s was to be the reference f o r RBE c a l c u l a t i o n s , but a l s o because of the c o n s i s t e n t dose r a t e and a v a i l a b i l i t y of the 6 0 C o radiotherapy u n i t . Two days p r i o r to each experiment, f r e s h l y t r y p s i n i z e d c e l l s were seeded i n t o each of two 75 cm 2 t i s s u e c u l t u r e f l a s k s a t 5-6 X 10 s c e l l s per f l a s k . V79 c e l l s were used f o r most experiments, but a few experiments were a l s o done w i t h CHO c e l l s . The growth medium i n the f l a s k s was re p l a c e d approximately 24 hours a f t e r p l a t i n g i n order to ensure optimal n u t r i t i o n a l c o n d i t i o n s f o r the growing c e l l s . On the day of the experiment, the c e l l s i n the two f l a s k s were t r y p s i n i z e d and pooled to form a s i n g l e parent suspension, the c o n c e n t r a t i o n of which was determined u s i n g a C o u l t e r Counter. C e l l s were then t r a n s f e r r e d to the ducks at a c o n c e n t r a t i o n of 2 X 10 s c e l l s / m l (20 ml t o t a l volume per duck). Growth medium without added bicarbonate b u f f e r was used i n these c e l l suspensions s i n c e the v e s s e l s would not be gassed w i t h a 5% C0 2-95% a i r mixture during the i r r a d i a t i o n procedure. In order to o b t a i n a dose response curve from a s i n g l e c e l l suspension, i r r a d i a t i o n s were performed i n dose increments, w i t h sample a l i q u o t s being removed from the duck a f t e r each increment. C e l l s were always exposed to a given r a d i a t i o n modality i n the same duck that was used i n the dosimetry measurements f o r that modality. A c o n t r o l (0 dose) a l i q u o t was removed from each duck immediately p r i o r to beginning the i r r a d i a t i o n s . For the generation of the low dose data, doses were administered i n approximately 0.4 Gy increments up to a t o t a l dose of e i t h e r 2.8 or 3.2 Gy. Exposure readings from the i o n i z a t i o n chamber were taken a f t e r each dose increment. To provide a comparison to e f f e c t s seen at more conventional r a d i o b i o l o g i c a l doses, h i g h dose data were a l s o 182 generated by subsequently a d m i n i s t e r i n g a 1.2 or 0.8 Gy increment (to b r i n g the t o t a l dose to -4 Gy), and then c o n t i n u i n g i n 2 Gy increments to a t o t a l dose of -10 Gy. High dose s u r v i v a l curves could then be generated from the 0, 2, 4, 6, 8, and 10 Gy samples. The volume of the a l i q u o t s removed a f t e r each dose increment was 0.5 ml, w i t h the exception of the 10 Gy dose p o i n t , f o r which a 3 ml a l i q u o t was taken. Because of the need to ensure accurate dose estimates, each sample a l i q u o t removed from the duck during the i r r a d i a t i o n procedure was replaced by an equal volume of f r e s h growth medium. This maintained the t o t a l volume of medium i n the duck at 20 ml. Since the a l i q u o t volume was s m a l l , a d d i t i o n of the f r e s h growth medium d i d not a f f e c t the c e l l c o n c e n t r a t i o n i n the suspension by more than about 25% from the f i r s t dose p o i n t to the l a s t . A l i q u o t s removed from the ducks were added to i n d i v i d u a l t e s t tubes c o n t a i n i n g d i l u t i o n volumes of medium. The d i l u t i o n volume f o r each dose p o i n t was such that the t o t a l volume of the f i n a l sample would be 10 ml a f t e r the i r r a d i a t e d a l i q u o t had been added. I r r a d i a t i o n s were performed at room temperature, one modality at a time. The order i n which the t e s t and reference samples were t r e a t e d was v a r i e d a t random. The second duck to be i r r a d i a t e d was kept at room temperature on a magnetic s t i r r e r (to keep the c e l l s i n suspension) while the f i r s t set of i r r a d i a t i o n s was performed. Each set of i r r a d i a t i o n s took 15-20 minutes. To i n h i b i t r e p a i r of r a d i a t i o n damage u n t i l the c e l l s could be p l a t e d and p r o p e r l y incubated at 37 °C, d i l u t i o n tubes c o n t a i n i n g samples from the f i r s t duck were placed on i c e w h i l e the second set of i r r a d i a t i o n s was performed. The second set of samples was a l s o p l a c e d on i c e a f t e r i r r a d i a t i o n f o r approximately the same amount of time as the f i r s t s e t . Thus, the only d i f f e r e n c e i n the way the two sets of samples were t r e a t e d was that the c e l l s of the second set had been kept i n 183 suspension at room temperature f o r 15-20 minutes longer than the f i r s t set p r i o r to treatment. In a d d i t i o n to the samples removed during the i r r a d i a t i o n procedure, a p o r t i o n of the c e l l p o p u l a t i o n used i n each experiment was reserved f o r DNA d i s t r i b u t i o n a n a l y s i s . The purpose of t h i s was to confirm that the c e l l p o p u l a t i o n used f o r each experiment was d i s t r i b u t e d evenly throughout the c e l l c y c l e ( i . e . t hat the p o p u l a t i o n was asynchronous). Synchrony i n the parent p o p u l a t i o n would a f f e c t the o v e r a l l r a d i a t i o n response of the c e l l s . The DNA d i s t r i b u t i o n a n a l y s i s was performed by c e n t r i f u g i n g samples of ~10 6 c e l l s f o r 6 minutes at 600 rpm. The supernatant was then poured o f f , and 1 ml of ethidium bromide (a q u a n t i t a t i v e DNA s t a i n ) was added to the r e s i d u a l c e l l p e l l e t . This mixture was vortexed v i g o r o u s l y to d i s t r i b u t e the s t a i n . The sample could then be analyzed w i t h a fluorescence a c t i v a t e d c e l l s o r t e r (FACS) to o b t a i n a histogram of DNA q u a n t i t y per c e l l f o r the p o p u l a t i o n . DNA d i s t r i b u t i o n s were inspected f o r the occurrence of c h a r a c t e r i s t i c s t r u c t u r e s produced by c e l l s i n the d i f f e r e n t stages of the c e l l c y c l e (see Figure 55) . For an asynchronously d i v i d i n g c e l l p o p u l a t i o n , a l l the c h a r a c t e r i s t i c s t r u c t u r e s i l l u s t r a t e d i n the f i g u r e were expected to be present. 7.2.3 P l a t i n g of Samples and Assessment of S u r v i v a l Once a l l i r r a d i a t i o n s had been completed, the samples from the d i l u t i o n tubes were p l a t e d i n t o e i t h e r 25 cm 2 t i s s u e c u l t u r e f l a s k s f o r the low dose assay, or i n t o 6 cm (10 cm f o r CHO experiments) diameter p e t r i dishes f o r the h i g h dose assay. For p l a t i n g purposes, the c e l l c o n c e n t r a t i o n i n each d i l u t i o n tube was estimated according to the s i z e and t o t a l dose f o r the a l i q u o t o r i g i n a l l y removed from the duck. P l a t i n g d e n s i t y f o r the low dose samples was 2500-3000 c e l l s per f l a s k f o r a l l dose 184 DNA content • Figure 55. D i s t r i b u t i o n of c e l l s i n the c e l l c y c l e . (a) the four phases of the c e l l c y c l e : G l (period of general p r o t e i n b i o s y n t h e s i s ) , S ( p e r i o d of DNA r e p l i c a t i o n ) , G2 ( p e r i o d during which s t r u c t u r e s a s s o c i a t e d w i t h m i t o s i s are s y n t h e s i z e d ) , M ( p e r i o d of m i t o s i s , or c e l l d i v i s i o n ) , (b) d i s t r i b u t i o n of DNA content i n a p o p u l a t i o n of asynchronously d i v i d i n g V79 c e l l s , as measured by a f l u o r e s c e n c e - a c t i v a t e d c e l l s o r t e r ( c e l l s were s t a i n e d w i t h ethidium bromide, a DNA-specific f l u o r e s c e n t dye). 185 p o i n t s . For the h i g h dose samples, the number of c e l l s p l a t e d per p e t r i d i s h was v a r i e d according to the expected s u r v i v a l r a t e i n such a way that there would be 150-200 (300-500 f o r CHO) s u r v i v o r s per p e t r i . Three r e p l i c a t e p e t r i s were p l a t e d per dose p o i n t f o r the h i g h dose assay, but only one f l a s k per dose p o i n t was p l a t e d f o r the low dose assay. The exact number of c e l l s p l a t e d per p e t r i d i s h f o r the h i g h dose samples was estimated from C o u l t e r counts performed d i r e c t l y on the d i l u t i o n tubes from which samples had been p l a t e d (counting was performed a f t e r a l l p l a t i n g had been completed). Low dose samples were incubated f o r 2 hours p r i o r to scanning. A l l f l a s k s were then kept at room temperature f o r the d u r a t i o n of the c e l l d e t e c t i o n / r e c o g n i t i o n phase of the experiment, which t y p i c a l l y l a s t e d about 6 hours. Scanning of the f l a s k s was performed according to the procedures discussed i n Chapters 4 and 5. The automated r e c o g n i t i o n procedure was used f o r the V79 c e l l s only, since r e c o g n i t i o n parameters had not been e s t a b l i s h e d f o r the CHO c e l l l i n e . Manual c l a s s i f i c a t i o n of objects i n the CHO samples was performed on the day a f t e r the experiment i n order to keep scanning times as short as p o s s i b l e . Since e s s e n t i a l l y a l l c e l l s i n an i r r a d i a t e d p o p u l a t i o n d i v i d e at l e a s t once at low to moderate doses (Grote et a l , 1981a; Brosing, 1983; E l k i n d et a l , 1963), t h i s delayed c l a s s i f i c a t i o n of c e l l s was not expected to have s i g n i f i c a n t e f f e c t s on the s u r v i v a l estimates. The o c c a s i o n a l V79 sample a l s o had to be manually c l a s s i f i e d due to hig h l e v e l s of d e b r i s , but t h i s was done immediately a f t e r scanning. In a d d i t i o n , spot checks of the V79 r e c o g n i t i o n a l g o r i t h m performance were done during each experiment by monitoring the f a l s e p o s i t i v e e r r o r r a t e i n 2-3 f l a s k s . F a l s e p o s i t i v e s o c c u r r i n g i n these f l a s k s were not c o r r e c t e d , 1 8 6 s i n c e i t was d e s i r a b l e to keep c e l l s e l e c t i o n c o n d i t i o n s as c o n s i s t e n t as p o s s i b l e between f l a s k s . Once scanning had been completed, a l l f l a s k s were returned to the incubator. They were then removed from the incubator only f o r b r i e f ( < 10-minute) i n t e r v a l s f o r colony scanning and/or s c o r i n g purposes. Colony f e a t u r e scores were c o l l e c t e d f o r the V79 samples at 2 days and 4 days post-treatment, p r i m a r i l y f o r the purposes of f u r t h e r t e s t i n g the automated colony s c o r i n g algorithms. The 3-point semi-automated f o c u s i n g method (with manual determination of the i d e a l focus s e t t i n g s at the 3 p o i n t s ) was used f o r these scans. Both V79 and CHO samples were al s o manually scored. P r e l i m i n a r y colony s c o r i n g was performed 4-5 days post-treatment, w i t h c o n f i r m a t i o n of questionable scores on day 6 or 7. The samples f o r the high dose assay were allowed a 7-day i n c u b a t i o n p e r i o d before being removed from the incubator and s t a i n e d . The number of c o l o n i e s per p e t r i d i s h was then counted without the a i d of a microscope (although some of the smaller c o l o n i e s were checked under the microscope to confirm that they d i d indeed c o n t a i n 50 c e l l s or more). 7.3 R e s u l t s and A n a l y s i s 7.3.1 Observations on Results f o r I n d i v i d u a l Experiments R e s u l t s obtained f o r i n d i v i d u a l low dose experiments were found to vary a f a i r amount between experiments, both i n terms of the o v e r a l l shapes of the s u r v i v a l curves, and i n the RBE's. F l u c t u a t i o n s seemed to be the g r e a t e s t f o r the 250 kV p experiments, f o r both the t e s t and c o n t r o l curves. Some v a r i a t i o n was a l s o seen i n the h i g h dose data, but the general shapes of the s u r v i v a l curves were r e l a t i v e l y c o n s i s t e n t , as were the RBE's c a l c u l a t e d at higher doses. 187 As a r e s u l t of the nature and extent of the f l u c t u a t i o n s seen i n the low dose data, the f i n a l determination of RBE and s u r v i v a l curve parameters was done on the combined data from a l l a v a i l a b l e experiments. In the case of the V79 data, only 6 of a t o t a l of 8 experiments per modality were used i n t h i s a n a l y s i s . The s e l e c t i o n of the experiments to be used was based on the q u a l i t y of the h i g h dose data, since problems w i t h f a u l t y C o u l t e r counts and b a c t e r i a l or fungal contamination of samples had been encountered i n some of the h i g h dose experiments. No s i m i l a r problems had occurred w i t h the low dose assay, but the same experiments as were chosen f o r the h i g h dose a n a l y s i s were used f o r the low dose a n a l y s i s . This provided a one-to-one correspondence ( i n terms of the c e l l p o p u l a t i o n used) between the RBE's obtained at h i g h and low doses. U n f o r t u n a t e l y , i t was not p o s s i b l e to achieve the same one-to-one correspondence f o r the CHO data, due to severe contamination problems i n s e v e r a l h i g h dose data s e t s . To compensate f o r t h i s , high dose experiments were repeated u n t i l the same number of s u c c e s s f u l experiments (3 per modality) had been performed at both h i g h and low doses. In one case, t h i s was done by generating 3 high dose curves (the 6 0Co 7-ray c o n t r o l curve plus two t e s t m o d a l i t i e s ) , i n a d d i t i o n to the usual 2 low dose curves, i n a s i n g l e experiment. 7.3.2 Methods of A n a l y s i s Two d i f f e r e n t approaches were used to c a l c u l a t e RBE's f o r the pooled s u r v i v a l data. The f i r s t assumed the v a l i d i t y of the l i n e a r - q u a d r a t i c equation f o r f i t t i n g s u r v i v a l data, a l l o w i n g RBE's to be c a l c u l a t e d from the d e r i v e d a-/? parameter values (Equations 7 .6 -7 .7 ) . The second approach r e l i e d on a simple method of i n t e r p o l a t i o n between data p o i n t s . The procedure f o l l o w e d i n t h i s instance was to perform a q u a d r a t i c l e a s t -squares f i t on corresponding sets of 3 neighbouring dose p o i n t s from the 188 t e s t and reference data. RBE's i n the v i c i n i t y of the middle p o i n t i n each set of 3 dose p o i n t s were c a l c u l a t e d from the d e r i v e d equations. The determination of a-B parameter values was somewhat more i n v o l v e d than the i n t e r p o l a t i o n method. While some p r e l i m i n a r y a n a l y s i s was done by performing a least-squares f i t to: In S (D) - In S (0) - aD - BD2 (7.8) m m (where D = dose and Sm(D) = measured s u r v i v a l at dose D (with 5^(0) commonly being r e f e r r e d to as the p l a t i n g e f f i c i e n c y ) ) , the f i n a l a n a l y s i s i n v o l v e d X2 m i n i m i z a t i o n of the equation: S(D) = e " a D - ^ 2 . (7.9) The x2 m i n i m i z a t i o n i n v o l v e d f i n d i n g the values of a and p t h a t gave the minimum value f o r X2 - 2 (l/a 1 2)[(S i-S(D 1 ) ] 2 (7.10) i where 5 i " W / V 0 > (7.11) and the a 2 were the u n c e r t a i n t i e s i n the data p o i n t s , S. . I t should be i r i noted t h a t the l e a s t squares f i t to Equation 7.8 was a l s o a x2 m i n i m i z a t i o n , except that In S , r a t h e r than S, was minimized, and a 2 was assumed to be u n i t y ( i n a d d i t i o n , In S m(0) was f i t t e d as a t h i r d parameter r a t h e r than being c a l c u l a t e d d i r e c t l y from the data at zero dose). 189 F i t t i n g (7.9) was much more d i f f i c u l t than f i t t i n g (7.8), s i n c e the s o l u t i o n c o u l d not be found a n a l y t i c a l l y . Nevertheless, because sampling e r r o r s i n S were governed by binomi a l s t a t i s t i c s , an approximately normal sampling e r r o r was expected f o r S r a t h e r than ln S. To o b t a i n the best p o s s i b l e estimates f o r a and /3, t h e r e f o r e (Ratkowsky, 1983; Bevington, 1969), the x2 m i n i m i z a t i o n was performed w i t h S as the dependent v a r i a b l e . Numerical methods had to be used to estimate the values a and that produced the minimum x2 value. Consequently, the number of parameters to be f i t t e d was kept to a minimum by not i n c l u d i n g the p l a t i n g e f f i c i e n c y as a t h i r d parameter. Rather, the data were normalized p r i o r to the a n a l y s i s u s i n g the mean measured p l a t i n g e f f i c i e n c y . This was b e l i e v e d to be j u s t i f i e d s i n c e enough data had been gathered to o b t a i n r e l a t i v e l y accurate estimates of S (0). Furthermore, zero-dose estimates could be based on the combined samples from the reference and t e s t curves f o r each r a d i a t i o n modality, s i n c e the c e l l s f o r both data sets had been t r e a t e d under e s s e n t i a l l y the same c o n d i t i o n s . This procedure l e d to a s i n g l e mean value of S (0) th a t was used to normalize both the t e s t and reference modality m data f o r each set of experiments. For the pooled data, the r e s u l t i n g value of S m(0) was w i t h i n ±0.01 of the S m(0) f o r the separate t e s t and reference curves, whether these l a t t e r p l a t i n g e f f i c i e n c i e s were obtained by averaging zero-dose data p o i n t s or by f i t t i n g ln 5^(0) as a t h i r d parameter i n Equation 7.8. A f t e r n o r m a l i z a t i o n of the data, the minimum x2 w a s sought by c a l c u l a t i n g x2 values on a g r i d of a and /3 values ( i n a f a s h i o n s i m i l a r to P a l c i c et a l , 1984). This technique not only allowed ( x 2 ) m i n to be found, but a l s o provided the necessary i n f o r m a t i o n to determine e r r o r estimates f o r a and /3. E r r o r s were determined by l o c a t i n g the (x 2) • +1 contours i n 190 parameter space. The a and B extrema of these contours give the standard d e v i a t i o n s f o r a and B, r e s p e c t i v e l y (Bevington, 1969). The m i n i m i z a t i o n of Equation (7.10) r e q u i r e d that estimates f o r be made. In general, i f a number of experiments have been done, these estimates can be obtained from the c a l c u l a t e d standard d e v i a t i o n s of the data p o i n t s . However, even though the RBE experiments had been repeated s e v e r a l times f o r each modality, much of the data c o u l d not be averaged because doses d i d not remain s u f f i c i e n t l y constant between experiments. An exception to t h i s was the 6 0Co 7-ray data. Values f o r at d i f f e r e n t s u r v i v a l l e v e l s were the r e f o r e based on standard d e v i a t i o n s c a l c u l a t e d from the c o b a l t data. These standard d e v i a t i o n s were obtained from the combined data of a l l the c o b a l t experiments, since t h i s would provide the best p o s s i b l e estimate of a Then, since the estimates were to be a p p l i e d to data from the other r a d i a t i o n m o d a l i t i e s , the dependence of cr on s u r v i v a l l e v e l was determined. This was done s e p a r a t e l y f o r the low and h i g h dose data, s i n c e d i f f e r e n t sources of e r r o r were i n v o l v e d . For the low dose assay, cr values to be used i n x2 minimizations were estimated from s t r a i g h t - l i n e f i t s of o /S versus S± f o r the 6 0 C o 7-ray data, as i l l u s t r a t e d i n Figure 56a. The aL values p l o t t e d i n the f i g u r e are the estimates made by f i n d i n g the standard d e v i a t i o n , cr i m, of the raw (unnormalized) c o b a l t data at each dose p o i n t , and then s c a l i n g according to: * i " °*JSm{0) (7.12) where S m(0) i s the mean p l a t i n g e f f i c i e n c y f o r a l l the c o b a l t experiments. The data shown i n Figure 56a represent 24 experiments f o r the V79 c e l l s and 9 experiments f o r the CHO c e l l s . Separate f i t s were performed f o r the V79 191 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 normalized survival IO"2 10"1 10° normalized survival F i g u r e 56. E s t i m a t i o n o f e x p e r i m e n t a l u n c e r t a i n t i e s a s s o c i a t e d w i t h the low and h i g h dose a s s a y s . E s t i m a t e s were ba s e d on the means and s t a n d a r d d e v i a t i o n s o f i n d i v i d u a l dose p o i n t s from r e p e a t e d e x p e r i m e n t s w i t h 6 0 C o 7 - r a y s . F r a c t i o n a l e r r o r i n the p l o t s r e p r e s e n t s the r a t i o : ( s t a n d a r d d e v i a t i o n ) / ( s u r v i v a l ) . (a) E x p e r i m e n t a l u n c e r t a i n t i e s a s s o c i a t e d w i t h the low dose a s s a y . A n a l y t i c a l a p p r o x i m a t i o n s t o t h e s e u n c e r t a i n t i e s were made w i t h l i n e a r f i t s t o the d a t a . (b) E x p e r i m e n t a l u n c e r t a i n t i e s a s s o c i a t e d w i t h the h i g h dose a s s a y . The a n a l y t i c a l a p p r o x i m a t i o n i s g i v e n by a q u a d r a t i c c u r v e . 192 and CHO data, s i n c e the experimental methods used w i t h the two c e l l l i n e s were s l i g h t l y d i f f e r e n t . For the h i g h dose data, the dependence of ai on s u r v i v a l was based on a p l o t of versus In St (Figure 56b) . A q u a d r a t i c curve was f i t t e d to these data p o i n t s to provide the estimates of as a f u n c t i o n of s u r v i v a l . The f i t was performed on the combined V79 (18 experiments) and CHO (8 experiments) data, s i n c e the dependence f o r both appeared to be s i m i l a r . 7.3.3 RBE Based on a-8 Parameters The pooled experimental r e s u l t s , along w i t h the c a l c u l a t e d l i n e a r -q u a d r a t i c f i t s , are shown f o r the low dose assay i n Figures 57 and 58, and f o r the h i g h dose assay i n Figures 59 and 60. The a-8 parameters corresponding to these f i g u r e s are l i s t e d i n Tables X and XI. P l a t i n g e f f i c i e n c i e s used to normalize each data set are a l s o l i s t e d . a values f o r both V79 and CHO c e l l s were q u i t e s i m i l a r , e s p e c i a l l y f o r the low dose r e s u l t s . B v a l u e s , on the other hand, d i f f e r e d c o n s i d e r a b l y due to the d i f f e r e n c e i n the s i z e of the s u r v i v a l curve "shoulder" f o r the two c e l l l i n e s . Furthermore, i t i s c l e a r that parameter values f o r the h i g h and low dose data were q u i t e d i f f e r e n t , e s p e c i a l l y f o r the q u a d r a t i c component, B. This v a r i a t i o n i n 8 between the low and h i g h dose f i t s may, however, have been p a r t i a l l y a t t r i b u t a b l e to the f a c t that 8 c o u l d not be a c c u r a t e l y estimated at low doses, where i t s c o n t r i b u t i o n to c e l l k i l l i n g was s m a l l . The low dose and high dose l i m i t s f o r the RBE, which are given by a/c*r and (8/Br)1/z, r e s p e c t i v e l y (Equation 7.7), are shown i n Table X I I . L i m i t s c a l c u l a t e d u s i n g both the low dose and the h i g h dose f i t s are i n c l u d e d , although 8 r a t i o s c a l c u l a t e d from low dose data are not l i k e l y to provide an accurate h i g h dose l i m i t (a s i m i l a r argument f o l l o w s f o r a r a t i o s c a l c u l a t e d from hi g h dose data). Despite any of the d i s c r e p a n c i e s 1 9 3 •*-60Co T-rays -^55 kVp X-rays > CD Co 7-rays •250 kVp X-rays 0.2 • A . ^ C O 7-rays 11 MeV electrons 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11 1 1 1 1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 dose (Gy) Figure 57. Low dose s u r v i v a l data f o r V79 c e l l s . Each p o i n t represents data from a s i n g l e f l a s k . a) c e l l s exposed to 6 0 C o 7-rays or exposed to 6 0Co 7-rays or 250 kV p X-rays. The normalized data (b) c e l l s 55 k V p X-rays. c) c e l l s exposed to 6 0Co 7-rays or 11 MeV e l e c t r o n s . were f i t t e d w i t h the l i n e a r - q u a d r a t i c equation using x 2 m i n i m i z a t i o n . a-/3 parameter values and p l a t i n g e f f i c i e n c i e s used f o r the f i t are l i s t e d i n Table X. 1 9 4 fl o • l—i o c d o - ^ C o 7-rays -•-55 kVp X-rays fl > 0.2 Co 7-rays -250 kVp X-rays o- 6 0Co 7-rays 11 MeV electrons 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 dose (Gy) Figure 58. Low dose s u r v i v a l data f o r CHO c e l l s . Each p o i n t represents data from a s i n g l e f l a s k . a) c e l l s exposed to 6 0Co 7-rays or 55 kVp X-rays. (b) c e l l s exposed to 6 0Co 7-rays or 250 kV p X-rays, c) c e l l s exposed to 6 0 C o 7-rays or 11 MeV e l e c t r o n s . The normalized data were f i t t e d w i t h the l i n e a r - q u a d r a t i c equation u s i n g \ 2 m i n i m i z a t i o n . a-8 parameter values and p l a t i n g e f f i c i e n c i e s used f o r the f i t are l i s t e d i n Table X. 195 10"N IO"2 4 o • r—i O •^ •60Co 7-rays -^-55 kVp X-rays P IO"2, P > IO-3-]- C o 7 _ r ay s -250 kVp X-rays IO"1, IO"2, •^ •60Co 7-rays 11 MeV electrons 0 8 10 12 dose (Gy) Figure 59. High dose s u r v i v a l data f o r V79 c e l l s . Each p o i n t represents data from a s i n g l e f l a s k . a) c e l l s exposed to 6 0 C o 7-rays or 55 kV p X-rays. (b) c e l l s exposed to 6 0Co 7-rays or 250 kV p X-rays. c) c e l l s exposed to 6 0Co 7-rays or 11 MeV e l e c t r o n s . The normalized data were f i t t e d w i t h the l i n e a r - q u a d r a t i c equation u s i n g x2 m i n i m i z a t i o n . Q-/3 parameter values and p l a t i n g e f f i c i e n c i e s used f o r the f i t are l i s t e d i n Table XI. 196 10° a o • r-H -»-> O u «+-( fl > cn IO" 3 10° IO"1 IO" 2 in-3 10° IO" 2 1 0 " 3 J (a) _; "0-60Co 7-rays ! -*-55 kVp X-rays (b) -o.60Co 7-rays ] -*-250 kVp X-rays l ^ - ^ (c) -c>-6()Co 7-rays ! -•-11 MeV electrons • i • i 1 i • i 0 2 4 6 8 1 i • 10 1 dose (Gy) Figure 60. High dose s u r v i v a l data f o r CHO c e l l s . Each p o i n t represents data from a s i n g l e f l a s k . a) c e l l s exposed to 6 0 C o 7-rays or 55 kV p X-rays. (b) c e l l s exposed to 6 0Co 7-rays or 250 kV p X-rays. c) c e l l s exposed to 6 0Co 7-rays or 11 MeV e l e c t r o n s . The normalized data were f i t t e d w i t h the l i n e a r - q u a d r a t i c equation u s i n g x2 m i n i m i z a t i o n . a-/3 parameter values and p l a t i n g e f f i c i e n c i e s used f o r f i t are l i s t e d i n Table XI. 197 T a b l e X. S u r v i v a l c u r v e p a r a m e t e r v a l u e s f o r low dose d a t a C e l l T e s t a r e f a test ^ r e f 8 test P l a t i n g L i n e R a d i a t i o n (Gy (Gy (Gy •2) (Gy 2) E f f i c i e n c y V79 55 kV 0 .20 0 .29 0 030 0 .041 0 85 X - r a y s ± 0 02 ± 0 03 ± 0 008 ± 0 011 ± 0 . 0 2 250 kV^ 0 23 0 .23 0 025 0 .051 0 87 ir P X - r a y s ± 0 02 ± 0 02 ± 0 008 ± 0 009 ± 0 . 0 3 11 MeV 0 19 0 .17 0 030 0 .039 0 85 e l e c t r o n s ± 0 02 ± 0 02 ± 0 008 ± 0 009 ± 0 . 0 3 CHO 55 kV 0 21 0 .30 0 056 0 .077 0 90 X - r a y s ± 0 03 ± 0 04 ± 0 011 ± 0 017 ± 0 . 0 2 250 kV„ 0 19 0 .20 0 049 0 .073 0 92 ir P X - r a y s ± 0 03 ± 0 03 ± 0 011 ± 0 013 ± 0 . 0 2 11 MeV 0 14 0 .14 0 068 0 .066 0 91 e l e c t r o n s ± 0 03 ± 0 03 ± 0 011 ± 0 011 ± 0 . 0 2 T a b l e XI. S u r v i v a l c u r v e p a r a m e t e r v a l u e s f o r h i g h dose d a t a C e l l T e s t Q r e f a test 0ref 0 test P l a t i n g L i n e R a d i a t i o n (Gy (cy (Gy •2) (Gy 2) E f f i c i e n c y V79 55 kV 0 27 0 .29 0 012 0 .019 0 81 X - r a y s ± 0 02 ± 0 02 ± 0 002 ± 0 002 ± 0 . 0 5 250 kV„ 0 24 0 .28 0 016 0 .018 0 84 ir P X - r a y s ± 0 02 ± 0 02 ± 0 002 ± 0 002 ± 0 . 0 7 11 MeV 0 25 0 .27 0 014 0 .013 0 86 e l e c t r o n s ± 0 02 ± 0 02 ± 0 002 ± 0 002 ± 0 . 0 9 CHO 55 kV 0 25 0 .30 0 030 0 045 0 81 X - r a y s ± 0 03 ± 0 03 ± 0 003 ± 0 003 ± 0 . 0 6 250 kV„ 0 20 0 .25 0 036 0 .043 0 83 ir P X - r a y s ± 0 03 ± 0 03 ± 0 003 ± 0 003 ± 0 . 0 5 11 MeV 0 21 0 .21 0 032 0 .032 0 85 e l e c t r o n s ± 0 03 ± 0 03 ± 0 003 ± 0 003 ± 0 . 0 9 Table XII. RBE l i m i t s f o r D •+ 0 and D -* » (D = dose), as c a l c u l a t e d from and B parameters V79 c e l l s CHO c e l l s (6 experiments/modality) (3 experiments/modality) Test " t e s t 8test 1 1 a t e s t 8test 1 7 R a d i a t i o n Q r e f Bref v. J Q r e f 3ref V. J Low Dose Data 55 kV p X-rays 1 5 ± 0.2 1.2 ± 0.2 1.4 ± 0.3 1.2 ± 0.2 250 k V p X-rays 1 0 ± 0.1 1.4 ± 0.3 1.1 ± 0.2 1.2 ± 0.1 11 MeV e l e c t r o n s 0 9 ± 0.1 1.1 ± 0.2 1.0 ± 0.3 1.0 ± 0.1 High Dose Data 55 kV p X-rays 1 1 ± 0.1 1.3 ± 0.1 1.2 ± 0.2 1.2 ± 0.07 250 k V p X-rays 1 2 ± 0.1 1.1 ± 0.09 1.2 ± 0.2 1.1 ± 0.06 11 MeV e l e c t r o n s 1 1 ± 0.1 1.0 ± 0.1 1.0 ± 0.2 1.0 ± 0.07 199 between the h i g h and low dose data, however, i t i s c l e a r t h a t none of the low dose RBE l i m i t s f o r X-rays were as high as p r e d i c t e d by the TDRA (nor were they as low as expected f o r the 11 MeV e l e c t r o n s ) ( c f . Table VII) . In an e f f o r t to o b t a i n more accurate estimates of a and /3 from a s i n g l e s u r v i v a l curve, an attempt was made to f i t these parameters to data sets c o n t a i n i n g values from both the high dose and low dose assays. To t h i s end, normalized low dose data f o r doses from 0 to -2 Gy ( i . e . the f i r s t 6 dose p o i n t s from each experiment) were combined w i t h normalized h i g h dose data from ~4 to -10 Gy ( i . e . the l a s t 4 dose p o i n t s from each experiment). The r e s u l t i n g a-/3 parameters, as w e l l as the low and high dose RBE l i m i t s , are l i s t e d i n Table X I I I . Once again, the a r a t i o s f o r the X-rays were lower than p r e d i c t e d by the TDRA, and were higher than expected f o r the 11 MeV e l e c t r o n s . A l l the /3 1 / 2 r a t i o s (with the p o s s i b l e exception of the CHO 55 kV p data) were, however, c l o s e to 1.0. 7.3.4 RBE Based on Interpolation Between Data Points The RBE's c a l c u l a t e d from i n t e r p o l a t i o n between data p o i n t s are l i s t e d i n Table XIV f o r both the high and the low dose assays. As i n the c a l c u l a t i o n of a and /3 valu e s , normalized s u r v i v a l data were used. The RBE's obtained at low doses using the i n t e r p o l a t i o n method were f o r the most p a r t s i m i l a r to those i n d i c a t e d by the a r a t i o s of Tables X I I and X I I I . High dose RBE's were s l i g h t l y higher than the corresponding /3 1 / 2 r a t i o s , although a dependence of RBE on dose was evident, e s p e c i a l l y f o r the X-ray data. The f a c t that the high dose RBE's were g e n e r a l l y not as clo s e to the value 1.0 as the corresponding /3 1 / 2 r a t i o s may have been due to the f a c t t h a t data were not obtained at s u f f i c i e n t l y h i g h doses to approximate the h i g h dose l i m i t ( i . e . D -*• «) . Table X I I I . S u r v i v a l curve parameter values and RBE l i m i t s f o r combined low dose-high dose f i t s Test R a d i a t i o n a r e f (Gy- *) a test (Gy- *) ^ r e f (Gy-2) &test (Gy-2) atest aref r &test { ?re£ \ 1 1 V79 cells 55 kV X-rays 0 23 ± 0. 01 0 .31 ± 0 01 0.016 ± 0.001 0.016 ± 0.002 1.3 ± 0. 07 1.0 ± 0. 07 250 kV X - r a y s P 0 25 ± 0. 01 0 .28 ± 0. 01 0.015 ± 0.001 0.018 ± 0.001 1.1 ± 0. 06 1.1 ± 0. 05 11 MeV el e c t r o n s 0 22 ± 0. 01 0 .21 + 0 01 0.019 + 0.002 0.021 ± 0.002 1.0 ± 0. 06 1.1 ± 0. 08 CHO cells 55 kV„ X-rays 0 25 ± 0. 02 0 .34 ± 0. 02 0.030 ± 0.002 0.041 ± 0.002 1.4 ± 0. 1 1.2 ± 0. 05 250 kV X - r a y s P 0 20 ± 0. 02 0 .24 + 0. 02 0.036 ± 0.002 0.043 ± 0.002 1.2 ± 0. 2 1.1 ± 0. 04 11 MeV el e c t r o n s 0 19 ± 0. 02 0 .18 ± 0 01 0.035 ± 0.002 0.035 ± 0.002 0.9 ± 0. 1 1.0 ± 0. 04 Table XIV. RBE's at d i f f e r e n t normalized s u r v i v a l l e v e l s , as obtained by i n t e r p o l a t i o n between data p o i n t s S u r v i v a l L e v e l RBE 55 kV p X-rays 250 kV p X-rays 11 MeV e l e c t r o n s V 7 9 cells 0 90 1 5 1 1 0 9 0 80 1 2 1 1 0 9 low dose 0 72 1 4 1 2 1 0 0 65 1 5 1 1 1 0 assay 0 56 1 3 1 1 1 0 0 48 1 3 1 2 1 0 0 50 1 2 1 2 1 0 high dose 0 25 1 1 1 1 1 0 0 10 1 2 1 1 1 0 assay 0 03 1 2 1 1 1 0 CHO cells 0 90 1 6 0 8 1 0 0 80 1 4 1 2 1 0 0 70 1 0 low dose 0 65 1 5 1 2 1 0 0 55 1 2 1 1 1 0 assay 0 48 1 2 1 1 1 0 0 40 1 3 1 1 1 0 0 50 1 3 1 2 1 0 high dose 0 20 1 2 1 1 1 0 0 05 1 2 1 1 1 0 assay 0 01 1 2 1 1 1 0 202 8. DISCUSSION 8 .1 C e l l Analyzer Performance 8.1.1 C e l l D e t e c t i o n and Recognition The c e l l d e t e c t i o n and r e c o g n i t i o n procedures f o r V79 c e l l s had been evaluated r e l a t i v e l y e x t e n s i v e l y during the course of t h e i r development, so a c e r t a i n l e v e l of performance was expected when they were used i n the RBE experiments. Moreover, because t h e i r performance was monitored during the course of these experiments, f u r t h e r e v a l u a t i o n s c o u l d be c a r r i e d out. O v e r a l l , i t was found that the c e l l d e t e c t i o n and r e c o g n i t i o n algorithms performed according to expectations i n terms of the number of c e l l s detected per f l a s k and the f a l s e p o s i t i v e e r r o r r a t e . For the V79 RBE experiments, the mean number of c e l l s assayed per f l a s k was 340 ± 35 (the e r r o r value i s the standard d e v i a t i o n ) f o r the e n t i r e set of experiments. For a 0.35 mm e x c l u s i o n d i s t a n c e , as was used i n the experiments, t h i s value i s 10-20% lower than what would be expected f o r 2500 c e l l s p l a t e d per f l a s k (with the presence of debris and the m i s c l a s s i f i c a t i o n of c e l l s by the r e c o g n i t i o n a l g o r i t h m taken i n t o account). The discrepancy could, however, be accounted f o r by f a c t o r s such as uneven d i s t r i b u t i o n of c e l l s and abnormal amounts of d e b r i s , which s i g n i f i c a n t l y diminished the s i z e of the p o s t - e x c l u s i o n data sets f o r some f l a s k s . Furthermore, up to 5% non-d e t e c t i o n of c e l l s was expected on the b a s i s of c a r e f u l e v a l u a t i o n of the performance of the c e l l d e t e c t i o n procedure. The number of c e l l s p l a t e d per f l a s k was a l s o only known approximately, due to p i p e t t i n g i n a c c u r a c i e s and to the f a c t that a c e r t a i n percentage of c e l l s i n a suspension may f a i l to adhere to the f l a s k surface. The mean number of CHO c e l l s assayed per f l a s k f o r the RBE experiments was 430 ±40. This value compares very w e l l w i t h the V79 203 average when i t i s considered that the CHO c e l l s were manually c l a s s i f i e d (15-25% of them were ther e f o r e not m i s c l a s s i f i e d as d e b r i s ) . Thus, while the performance of the c e l l d e t e c t i o n procedure could not be evaluated d i r e c t l y during the course of the experiments, the c e l l numbers obtained suggest t h a t i t was performing as expected. Furthermore, i n those instances where c e l l s detected i n the f l a s k s were manually r e v i s i t e d , the incidence of two c e l l s being too c l o s e together ( i n d i c a t i n g that one of the c e l l s had not been detected) was very low. The performance of the r e c o g n i t i o n a l g o r i t h m was checked f o r 1-3 f l a s k s i n each experiment w i t h V79 c e l l s . These checks y i e l d e d a f a l s e p o s i t i v e e r r o r r a t e of 2.0+1.2%, which i s i d e n t i c a l to the f a l s e p o s i t i v e e r r o r s obtained during independent ev a l u a t i o n s of the r e c o g n i t i o n a l g o r i t h m ( c f . Table V). 8.1.2 C l a s s i f i c a t i o n of Colonies While the automated colony s c o r i n g algorithms were not used i n the d e t a i l e d a n a l y s i s of any of the RBE experiments, they too were f u r t h e r t e s t e d u s i n g data c o l l e c t e d during the course of these experiments. Of the two methods developed i n Chapter 6, the colony screening method ( S e c t i o n 6.5.2) gave the best s u r v i v a l estimates. However, because of the inaccurate f o c u s i n g methods used i n the gathering of the data, a 500-2500 " t o t a l area" screening i n t e r v a l was necessary to achieve acceptable l e v e l s of performance. With t h i s screening i n t e r v a l , s u r v i v a l estimates deviated by no more than AS = ±0.03 from manual measurements, w i t h an average d e v i a t i o n over a l l data sets of ±0.01. This e r r o r r a t e was near the achievable minimum, since approximately 1.5% of the c o l o n i e s i n a t y p i c a l sample co u l d not be scored manually w i t h any degree of c e r t a i n t y due to t h e i r s i z e (-50 c e l l s ) at 6-7 days p o s t - i r r a d i a t i o n . However, as expected, 204 manual s c o r i n g had to be r e l i e d upon f o r 33% of the c o l o n i e s i n order to achieve t h i s l e v e l of performance. The a-p values that were obtained f o r the pooled V79 low dose experiments u s i n g the screening method are compared to the values obtained by s t r i c t l y manual s c o r i n g i n Table XV ( a l l 8 experiments per t e s t modality were i n c l u d e d i n the a n a l y s i s ) . The parameter values obtained u s i n g the screening method were w e l l w i t h i n experimental e r r o r of the values d e r i v e d from the manual c l a s s i f i c a t i o n s . In c o n t r a s t to the screening method, the automated s c o r i n g method based on the use of colony growth r a t e i n f o r m a t i o n ( S e c t i o n 6.5.3) d i d not perform w e l l on the experimental RBE data. In every experiment, at l e a s t one data p o i n t deviated by 10% or more from the manually determined s u r v i v a l . As w i t h the p r e l i m i n a r y colony f e a t u r e data discussed i n Chapter 6, these e r r o r s were thought to be l a r g e l y a t t r i b u t a b l e to i n c o n s i s t e n t f o c u s i n g . I t t h e r e f o r e appears t h a t improved colony f o c u s i n g procedures would make a v a l u a b l e c o n t r i b u t i o n to the c a p a b i l i t i e s of the C e l l Analyzer i n terms of the gathering of p o p u l a t i o n d e n s i t y data. 8.1.3 Comparison to Previously Used Low Dose Surv i v a l Assays As discussed i n S e c t i o n 1.1.3, low dose s u r v i v a l s t u d i e s have p r e v i o u s l y been made usin g manual or semi-automated methods. These methods were extremely l a b o u r - i n t e n s i v e and time-consuming. T h e i r automation through the development of c e l l l o c a t i n g and r e v i s i t i n g procedures f o r the C e l l Analyzer y i e l d e d a considerable improvement not only i n the amount of data t h a t c o u l d be c o l l e c t e d i n a given p e r i o d of time, but a l s o i n the degree of s t r a i n experienced by the experimenter. Using the procedures developed during the course of t h i s t h e s i s , 300-400 i s o l a t e d c e l l s could be l o c a t e d i n a s i n g l e t i s s u e c u l t u r e v e s s e l 205 Table X V . Comparison of s u r v i v a l curve parameter values obtained f o r V79 low dose data w i t h manual s c o r i n g of s u r v i v o r s and w i t h s c o r i n g u s i n g the colony screening method Test R a d i a t i o n Parameter Manual Scoring Colony Screening 55 kV p X-rays 0.85 ± 0 03 0.85 ± 0.03 " r e f ^ y " 1 ) 0.20 ± 0 02 0.19 + 0.02 " t e s t ( C y - 1 ) 0.28 + 0 02 0.27 ± 0.02 B r e f (Gy-*) 0.030 ± 0 007 0.029 ± 0.007 Btest ( G y 2 ) 0.04 ± 0 01 0.04 ± 0.01 a t e s t ^ a r e f 1.4 ± 0 2 1.4 ± 0.2 ^ t e s t ^ r e f ) 1 7 2 1.2 ± 0 2 1.2 ± 0.2 250 kV p X-rays S o 0.87 ± 0 03 0.87 ± 0.03 " r e f (GY" 1) 0.22 ± 0 02 0.22 ± 0.02 " t e s t (Gy" 1) 0.24 + 0 02 0.24 + 0.02 Pref (Gy" 2) 0.031 ± 0 007 0.030 + 0.007 Ptest (Gy' 2) 0.049 ± 0 008 0.047 + 0.008 a t e s t / a r e f 1.1 ± 0 1 1.1 ± 0.1 ^test/Pref^/2 1.3 + 0 2 1.3 ± 0.2 11 MeV e l e c t r o n s So 0.85 + 0 04 0.85 ± 0.04 " r e f (Gy" 1) 0.20 ± 0 02 0.20 + 0.02 " t e s t (Gy" 1) 0.17 ± 0 02 0.17 + 0.02 f3ref (Gy" 2) 0.030 ± 0 007 0.027 ± 0.007 Ptest (Gy' 2) 0.042 ± 0 008 0.039 ± 0.008 " t e s t / " r e f 0.9 + 0 1 0.9 ± 0.1 ( ^ e s t ^ r e f ) l / 2 1.2 + 0 2 1.2 ± 0.2 SQ i s the p l a t i n g e f f i c i e n c y 206 i n the course of 15-20 minutes. By comparison, only 100-200 c e l l s c ould be l o c a t e d i n a s i n g l e v e s s e l by manual methods (Brosing, 1983), and even t h i s r e q u i r e d a scanning time of approximately 30-40 minutes per v e s s e l . Semi-automated methods employing a computerized microscope stage combined w i t h manual c e l l s e l e c t i o n ( P a l c i c et a l , 1983; Faddegon, 1983) shortened the scanning time to approximately 20 minutes per v e s s e l , but the number of c e l l s t hat c o u l d be s e l e c t e d per v e s s e l without a p p r e c i a b l e r i s k of d u p l i c a t i o n remained about the same. Because a l a r g e r number of c e l l s could be assayed per experiment, the sampling e r r o r a s s o c i a t e d w i t h low dose s u r v i v a l data gathered u s i n g the C e l l Analyzer was lower than that achievable by manual or semi-automated methods. This was borne out by smaller variances a s s o c i a t e d w i t h data c o l l e c t e d at i n d i v i d u a l dose p o i n t s ( r e s u l t s from a set of 3 i d e n t i c a l experiments w i t h CHO c e l l s exposed to 250 kV p X-rays i n Faddegon (1983) cou l d be compared w i t h a s i m i l a r set of 3 CHO c e l l experiments presented i n t h i s t h e s i s ) . Consequently, a, parameter estimates made from the manual and semi-automated s t u d i e s were a l s o l e s s accurate than those obtained us i n g the C e l l Analyzer (the percentage e r r o r s i n the parameter estimates i n Faddegon (1983) were up to a f a c t o r of 3 higher than those obtained i n t h i s t h e s i s ) . Estimates of RBE's (see Faddegon, 1983; B r o s i n g , 1983) were s i m i l a r l y a f f e c t e d . I t should, however, be noted that the RBE's (or enhancement r a t i o s , as they are r e f e r r e d to when the e f f e c t s of r a d i a t i o n s e n s i t i z e r s are being measured) i n these e a r l i e r low dose s t u d i e s were c a l c u l a t e d d i f f e r e n t l y than i n t h i s t h e s i s , so a d i r e c t comparison could not e a s i l y be made. 207 8.1.4 Assessment of E r r o r s A s s o c i a t e d w i t h the Low Dose Assay The purpose i n developing a low dose assay u s i n g the C e l l Analyzer was to o b t a i n data at h i g h s u r v i v a l l e v e l s w i t h greater accuracy than i s p o s s i b l e by conventional counting and p l a t i n g techniques, or by l e s s s o p h i s t i c a t e d methods of c e l l s e l e c t i o n and follow-up. As seen i n the previous s e c t i o n , an increase i n the number of c e l l s assayed per experiment by automation of the c e l l s e l e c t i o n procedure could l e a d to a s i g n i f i c a n t r e d u c t i o n i n the experimental e r r o r . I r r e s p e c t i v e of the method used, however, experimental accuracy i s always a f f e c t e d by sampling e r r o r , procedural e r r o r s , and b i o l o g i c a l v a r i a b i l i t y . While sampling e r r o r s can be determined from s t a t i s t i c a l c o n s i d e r a t i o n s and procedural e r r o r s can be estimated from c o n t r o l s t u d i e s , b i o l o g i c a l v a r i a b i l i t y must be assessed from r e p e t i t i o n of experiments. A l l the above procedures were c a r r i e d out f o r the low dose assay during the course of t h i s t h e s i s , so a r e l a t i v e l y complete assessment of the e r r o r s a s s o c i a t e d w i t h t h i s assay c o u l d be made. Two main procedural e r r o r s c o u l d be i d e n t i f i e d w i t h the low dose assay. These were the f a l s e p o s i t i v e e r r o r r a t e f o r c e l l r e c o g n i t i o n , and the i n c l u s i o n of c e l l doublets i n the data s e t . Estimates of the magnitudes of both these q u a n t i t i e s were made from c o n t r o l s t u d i e s , and, i n the case of the f a l s e p o s i t i v e e r r o r , were v e r i f i e d during the course of the RBE experiments. Their e f f e c t on s u r v i v a l estimates c o u l d be determined by examining how they would a f f e c t the number of s u r v i v i n g c o l o n i e s i n a sample w i t h a "t r u e " s u r v i v a l p r o b a b i l i t y , S. For a f r a c t i o n a l f a l s e p o s i t i v e e r r o r r a t e of f, s u r v i v a l estimates would be a f f e c t e d according t o: S t - S ( 1 - f ) ( 8 . 1 ) 208 where S f i s the measured s u r v i v a l l e v e l f o r the sample. The e f f e c t of doublets on s u r v i v a l measurements co u l d a l s o be estimated, subject to c e r t a i n s i m p l i f y i n g assumptions. The f i r s t of these approximations was that a n e g l i g i b l e number of c e l l s d i v i d e d u r ing the two hour i n c u b a t i o n p e r i o d before scanning ( t h i s was f e l t to be j u s t i f i e d s ince a c e l l p o p u l a t i o n t y p i c a l l y m u l t i p l i e s at a diminished r a t e f o r a p e r i o d of time a f t e r t r y p s i n i z a t i o n and p l a t i n g ) , and the second was that a l l c e l l s i n the p o p u l a t i o n have equal r a d i o s e n s i t i v i t y , i n c l u d i n g each i n d i v i d u a l member of a doublet. The e f f e c t of doublets on the o v e r a l l s u r v i v a l could then be determined u s i n g the binomi a l d i s t r i b u t i o n to f i n d the p r o b a b i l i t y of one, both, or none of the members i n a doublet s u r v i v i n g . The p r o b a b i l i t y , P , of a doublet being scored as a s u r v i v o r was then the p r o b a b i l i t y that e i t h e r or both members s u r v i v e : P d = S (2-S) (8.2) where S i s again the "tr u e " s u r v i v a l l e v e l . I f d i s the f r a c t i o n of doublets i n the data s e t , NS(l-d)+NS(2-S)d S J S+Sd-S2d (8.3) d JV where S d i s the measured s u r v i v i n g f r a c t i o n and N i s the number of ( c e l l s + doublets) i n the data s et. Combining the e f f e c t s of the two sources of e r r o r by s e t t i n g S = Sd i n Equation 8.1, S = (l-f)(S+Sd-S2d) m (8.4) 209 where i s the measured s u r v i v a l t a k i n g both sources of e r r o r i n t o account. Since the f r a c t i o n of f a l s e p o s i t i v e s and the f r a c t i o n of doublets were sub j e c t to some v a r i a b i l i t y , they c o n t r i b u t e d both to the e r r o r i n e s t i m a t i n g the s u r v i v a l and to the v a r i a b i l i t y seen when experiments were repeated. T h e i r expected c o n t r i b u t i o n to the t o t a l v a r i a n c e , a 2 f o r each measured dose p o i n t on a s u r v i v a l curve c o u l d be c a l c u l a t e d according to: CTsm2 = °s2(dSm/dS)t + atHdSm/df)* + a^(dSJdd)^ (8.5) (Bevington, 1969) where a f 2 and a 2 ave the variances a s s o c i a t e d w i t h the f a l s e p o s i t i v e e r r o r and the f r a c t i o n of doublets, r e s p e c t i v e l y , and a s 2 = S(l-S)/Nt i s the binom i a l sampling v a r i a n c e . N i s the t o t a l number of c e l l s assayed per dose p o i n t i n a s i n g l e experiment ( i n c l u d i n g i n d i v i d u a l members of c e l l d o u b l e t s ) : Nt - N (1-f) (1+d). (8.6) The expected variance c a l c u l a t e d according to Equation 8.5 could be compared to the var i a n c e observed i n a s e r i e s of experiments. In t h i s way, the magnitude of the b i o l o g i c a l v a r i a b i l i t y c ould be estimated. The 6 0 C o 7-ray data gathered as c o n t r o l curves i n the RBE experiments could conven i e n t l y be used, since a l a r g e number of them were generated and, furthermore, s u r v i v a l data could be averaged at s p e c i f i c dose p o i n t s . A comparison of the measured s u r v i v a l (mean and v a r i a n c e ) , the "t r u e " s u r v i v a l (from Equation 8.4), and the expected v a r i a n c e due to sampling and procedural e r r o r s (Equation 8.5) i s shown i n Table XVI f o r both the V79 and T a b l e X V I . Summary o f s t a n d a r d d e v i a t i o n s f o r t h e 6 0 C o 7 - r a y low dos d a t a Dose (Gy) Measured S u r v i v a l Measured °s "True" S u r v i v a l E x p e c t e d as E x p e c t e d a g Measured as V79 cells 0 0 860 0 033 0 871 0 020 0 61 0.425 ± 0 003 0 783 0 036 0 789 0 023 0 64 0.849 ± 0 006 0 700 0 035 0 702 0 025 0 72 1.28 + 0 01 0 637 0 031 0 636 0 026 0 85 1.70 + 0 01 0 547 0 042 0 544 0 027 0 64 2.13 + 0 02 0 479 0 045 0 .474 0 027 0 60 2.55 ± 0 02 0 410 0 031 0 .404 0 027 0 86 3.00 ± 0 02 0 365 0 045 0 .358 0 026 0 58 3.384 + 0 007 0 312 0 033 0 .305 0 025 0 75 CHO cells 0 0 908 0 017 0 903 0 013 0 77 0.415 ± 0 002 0 844 0 027 0 .836 0 017 0 62 0.829 + 0 004 0 753 0 033 0 .742 0 020 0 61 1.246 ± 0 008 0 673 0 043 0 .659 0 022 0 52 1.661 + 0 008 0 565 0 039 0 550 0 024 0 60 2.07 ± 0 01 0 484 0 055 0 469 0 024 0 43 2.49 ± 0 01 0 397 0 052 0 382 0 024 0 45 2.91 ± 0 02 0 329 0 038 0 316 0 023 0 59 3.32 + 0 02 0 295 0 030 0 283 0 022 0 73 211 the CHO low dose 6 0 C o 7-ray data. Sample means and variances were c a l c u l a t e d from a t o t a l of 24 experiments f o r the V79 c e l l l i n e and 9 experiments f o r the CHO c e l l l i n e . In the c a l c u l a t i o n s , the mean measured s u r v i v a l was taken to be an accurate estimate of S , a reasonable m assumption due to the la r g e number of experiments performed. The f a l s e p o s i t i v e e r r o r ( f ± a£) was 0.020 ±0.012 (from S e c t i o n 8.1.1), the f r a c t i o n of doublets (d ± a d) was 0.06 ±0.02 (estimated from c o n t r o l experiments), and the value of N was 340 f o r the V79 experiments and 428 f o r the CHO experiments ( S e c t i o n 8.1.1). The comparisons were made f o r unnormalized s u r v i v a l data, s i n c e the zero-dose p o i n t was subj e c t to the same sources of e r r o r as the other dose p o i n t s . As seen i n Table XVI, the v a r i a b i l i t y due to sampling and procedural e r r o r s ("expected &s") accounts f o r , on average, 50-80% of the t o t a l v a r i a b i l i t y . In f a c t , much of the expected v a r i a b i l i t y ( i . e . > 95%) was due to the sampling e r r o r alone, i n d i c a t i n g that the procedural e r r o r s had been c o n t r o l l e d to a high degree. Likewise, a d d i t i o n a l sources of procedural e r r o r s that were not inc l u d e d i n the estimates of Table XVI were e q u a l l y s m a l l . These in c l u d e d f l u c t u a t i o n s i n the administered dose (±0.6-0.7% f o r the 6 0 C o 7-rays), as w e l l as u n c e r t a i n t i e s i n the s c o r i n g of s u r v i v o r s (±1.5%). Indeed, a rough estimate of the c o n t r i b u t i o n due to u n c e r t a i n t i e s i n s c o r i n g s u r v i v o r s r e s u l t e d i n a n e g l i g i b l e increase i n the "expected" value of a„ 2 . Since a l a r g e p a r t of the expected v a r i a b i l i t y was due to the sampling e r r o r , s i g n i f i c a n t improvements i n N would be expected to reduce the o v e r a l l s c a t t e r i n the data. The number of experiments that would have to be performed i n order to achieve a c e r t a i n confidence i n t e r v a l f o r the s u r v i v a l estimates would a l s o decrease. The most l i k e l y means of i n c r e a s i n g N would be to reduce the m i s c l a s s i f i c a t i o n r a t e f o r c e l l s i n the 212 r e c o g n i t i o n a l g o r i t h m , but -a much more e f f e c t i v e improvement would be made by i n c r e a s i n g the number of f l a s k s scanned per experiment. This would be p o s s i b l e even w i t h the present system and algorithms, e s p e c i a l l y i f more than one C e l l Analyzer were a v a i l a b l e f o r simultaneous use. S i g n i f i c a n t gains would a l s o be made by reducing the scanning time per f l a s k , or, a l t e r n a t i v e l y , by developing d e t e c t i o n and r e c o g n i t i o n algorithms that would a l l o w the s e l e c t i o n of p o p u l a t i o n samples from f l a s k s t h a t have been incubated f o r longer than 2 hours a f t e r treatment. The l a t t e r method would have the advantage of not r e q u i r i n g f l a s k s to be kept at room temperature f o r the d u r a t i o n of the c e l l d e t e c t i o n p o r t i o n of the experimental procedure. C e l l d e t e c t i o n and r e c o g n i t i o n algorithms f o r such a task would have to be able to c o r r e c t l y s e l e c t both s i n g l e c e l l s and m i c r o c o l o n i e s of 2-4 c e l l s f o r the data set. U n f o r t u n a t e l y , none of the doublets present i n the p o p u l a t i o n at the time of i r r a d i a t i o n would be i d e n t i f i a b l e at the time of scanning, so t h i s source of e r r o r would be unavoidable. A l l o w i n g a longer i n c u b a t i o n time before s e l e c t i o n of the data sets would a l s o reduce e r r o r s caused by p o o r l y adhered c e l l s detaching from the f l a s k surface a f t e r they had already been detected and i n c l u d e d i n the data set . In some V79 experiments, up to 3-4% of the c e l l s would detach between the time of scanning and subsequent r e v i s i t s 1-2 days l a t e r (the problem d i d not a r i s e i n the CHO experiments because manual c e l l r e c o g n i t i o n was performed 1 day a f t e r scanning). The cause of the c e l l detachment has not been determined, but i t s occurrence d i d place a d d i t i o n a l l i m i t a t i o n s on the accuracy of p l a t i n g e f f i c i e n c y estimates, and may a l s o have c o n t r i b u t e d to the o v e r a l l v a r i a n c e i n the data. Indeed, t h i s may have been the reason why the v a r i a n c e i n the p l a t i n g e f f i c i e n c y was much higher f o r the V79 than f o r the CHO experiments ( c f . Table XVI). Since the p r o p o r t i o n of s u r v i v o r s versus non-survivors i s not known f o r the p o p u l a t i o n of detaching c e l l s , 213 the c o n t r i b u t i o n of t h i s phenomenon to the e r r o r i n the s u r v i v a l estimates i s d i f f i c u l t to determine. ( I t should be noted, however, that the f r a c t i o n of detached c e l l s per f l a s k d i d not appear to be dose dependent) . In the f u t u r e , i f c e l l d e t e c t i o n and r e c o g n i t i o n continues to be c a r r i e d out on the day of p l a t i n g , i t may be u s e f u l to augment these procedures w i t h an automated r e v i s i t of the s e l e c t e d c e l l l o c a t i o n s 1-2 days a f t e r p l a t i n g . Locations i n which c e l l s have completely disappeared c o u l d then be e l i m i n a t e d from f u r t h e r c o n s i d e r a t i o n . Algorithms would have to be developed to d i s t i n g u i s h "blank" l o c a t i o n s from those c o n t a i n i n g s i n g l e c e l l s or m i c r o c o l o n i e s , w i t h the c o n s i d e r a t i o n that pieces of debris i n "blank" l o c a t i o n s could produce a detectable s i g n a l . I f automated colony s c o r i n g u s i n g the d i s c r i m i n a n t f u n c t i o n method of S e c t i o n 6.5.4 i s als o improved to the p o i n t of p r a c t i c a l a p p l i c a b i l i t y , the day 2 colony scan co u l d l i k e l y be combined w i t h the abovementioned search f o r "blank" l o c a t i o n s , and p o s s i b l y a l s o w i t h a c o n f i r m a t i o n of the c e l l c l a s s i f i c a t i o n s made on the day of scanning (as p r e v i o u s l y discussed i n Se c t i o n 6 .5 .4 ) . Even w i t h s i g n i f i c a n t improvements i n sampling e r r o r s through increases iV, i t i s evident from Table XVI that a f a i r amount of the v a r i a b i l i t y seen i n the data i s due to u n c o n t r o l l e d f a c t o r s c o n t r i b u t i n g to the s o - c a l l e d " b i o l o g i c a l v a r i a b i l i t y " between c e l l populations t r e a t e d on d i f f e r e n t days. A d d i t i o n a l evidence of t h i s source of v a r i a b i l i t y can be seen when some of the a and (i parameter values f o r the 6 0 C o 7-ray data used as reference curves f o r the RBE experiments are compared. One c l e a r example of t h i s can be seen i n the CHO low dose experiments (Table X) , where <*ref and Pref f o r the 11 MeV e l e c t r o n experiments d i f f e r markedly from &ref and Pref f o r the other two m o d a l i t i e s . P o s s i b l e sources of these u n c o n t r o l l e d p e r t u r b a t i o n s include p h y s i c a l damage during p i p e t t i n g , damage 214 c a u s e d by t r y p s i n i z a t i o n , s l i g h t changes i n the c o m p o s i t i o n o f t h e growth medium, f l u c t u a t i o n s i n t h e t e m p e r a t u r e and pH t o w h i c h t h e c e l l s a r e exposed, t h e p r e s e n c e o f t o x i c c o n t a m i n a n t s , t h e d i s t r i b u t i o n o f c e l l s i n the c e l l c y c l e , and random m u t a t i o n i n the c e l l p o p u l a t i o n . W h i l e most o f t h e s e f a c t o r s c a n be c o n t r o l l e d t o a c e r t a i n d e g r e e , t h e i r combined e f f e c t i s o f t e n l a r g e enough t o cause a p p r e c i a b l e d i f f e r e n c e s i n c e l l u l a r r e s p o n s e f r o m e x p e r i m e n t t o e x p e r i m e n t . T h i s s o u r c e o f v a r i a t i o n must al w a y s be c o n s i d e r e d when the e f f e c t s o f damaging a g e n t s on c e l l s a r e t e s t e d , r e g a r d l e s s o f t h e a c c u r a c y o f t h e e x p e r i m e n t a l methods u s e d . 8.2 RBE E x p e r i m e n t s 8.2.1 D e t e r m i n a t i o n o f a-8 and RBE V a l u e s RBE v a l u e s o b t a i n e d u s i n g b o t h l o c a l i z e d i n t e r p o l a t i o n between d a t a p o i n t s and l i n e a r - q u a d r a t i c f i t s t o c o m p l e t e d a t a s e t s were comparable t o each o t h e r f o r a l l the r a d i a t i o n m o d a l i t i e s t e s t e d . Thus, i t appears t h a t the l i n e a r - q u a d r a t i c s u r v i v a l model can be r e l i e d upon t o o b t a i n r e a s o n a b l e e s t i m a t e s o f RBE, a t l e a s t f o r low LET r a d i a t i o n s . However, i t was a p p a r e n t t h a t some c a r e had t o be t a k e n i n t h e i n t e r p r e t a t i o n o f v a l u e s o f the q u a d r a t i c component, 8, e s p e c i a l l y i f i t was o b t a i n e d s t r i c t l y from t h e low dose d a t a . I n t h e low dose range, 8 makes o n l y a s m a l l c o n t r i b u t i o n t o t o t a l c e l l k i l l i n g , e s p e c i a l l y f o r V79 c e l l s . T h i s c a n be seen by e x a m i n i n g t h e a/6 r a t i o , w h i c h g i v e s the dose a t w h i c h the l i n e a r and q u a d r a t i c c o n t r i b u t i o n s t o c e l l k i l l i n g a r e e q u a l . F o r V79 c e l l s exposed t o 6 0 C o 7-rays, a/8 = 10-15 Gy, w h i c h i s even beyond t h e range o f most h i g h dose d a t a . Even f o r CHO c e l l s , w h i c h t e n d t o e x h i b i t a l a r g e r q u a d r a t i c component o f c e l l k i l l i n g , a/8 « 5 Gy o r more. 215 Thus, i t appears that the generation of h i g h dose data to accompany any low dose experiments would be u s e f u l f o r the determination of s u r v i v a l curve parameters. However, while i t i s suggested that d i f f e r e n c e s i n a and /3 between h i g h and low dose f i t s may be a t t r i b u t a b l e to d i f f i c u l t i e s i n f i t t i n g s u r v i v a l data w i t h the necessary accuracy, the p o s s i b i l i t y e x i s t s that these two parameters are not s u f f i c i e n t to provide a proper f i t to mammalian c e l l s u r v i v a l data ( a l t e r n a t i v e l y , a and/or /3 values may be dose-dependent). Indeed, there has been some evidence from other i n v e s t i g a t o r s (Skarsgard et a l , 1987 and 1989) that c e l l s u r v i v a l data generated from asynchronous c e l l populations show some substructure i n the low dose range. I f t h i s i s the case, s u r v i v a l curve parameters obtained from hi g h and low dose f i t s would be expected to be d i f f e r e n t . Some evidence that a-/3 values may not be constant w i t h dose can be i n f e r r e d from the data obtained f o r t h i s t h e s i s . For example, when the combined 6 0 C o 7-ray data f o r a l l experiments were examined, d i s t i n c t p a t t e r n s of change i n a and /3 could be seen as a f u n c t i o n of the dose range used to estimate them. The r e s u l t s of t h i s a n a l y s i s are summarized i n Table XVII. For both V79 and CHO c e l l s , a increased s t e a d i l y w i t h i n c r e a s i n g dose range, while /S decreased. Furthermore, values of Xj2 = (x 2) m i n/(N-2) (where iV i s the t o t a l number of data p o i n t s used f o r the f i t ) , which i s a measure of the "goodness of f i t " (Bevington, 1969) , were r e l a t i v e l y constant and c l o s e to the value of 1.0 (which i n d i c a t e s a good f i t ) f o r a l l examples shown i n the t a b l e . Nevertheless, i t i s c l e a r that more extensive s u r v i v a l data than were obtained f o r t h i s t h e s i s would be r e q u i r e d to r e s o l v e t h i s i s s u e , e s p e c i a l l y when i t i s considered that the xj1 values obtained f o r the combined low dose/high dose f i t s of Table X I I I were not a p p r e c i a b l y d i f f e r e n t than those obtained f o r the i n d i v i d u a l h i g h and low dose f i t s of Tables X and XI. Table XVII. V a r i a t i o n of a-3 estimates w i t h dose range used f o r f i t (6oco 7-ray s u r v i v a l data) Dose Range Source of Data a X 2 V79 c e l l s 0 - -2 Gy low dose assay 0 19 0 039 1.1 0 - -3 Gy low dose assay 0 21 0 030 1.0 0 - -10 Gy high dose assay 0 25 0 014 1.0 CHO c e l l s 0 - -2 Gy low dose assay 0 14 0 082 0.9 0 - -3 Gy low dose assay 0 17 0 060 1.0 0 - -10 Gy high dose assay 0 21 0 034 0.9 217 8.2.2 R e l a t i o n s h i p of Results to Models of R a d i a t i o n A c t i o n Despite the d i f f i c u l t i e s i n determining RBE values w i t h a h i g h degree of accuracy, the experiments d i d r e v e a l that the s i g n i f i c a n t increase i n RBE seen w i t h decreasing dose f o r X-rays versus 6 0 C o 7-rays usi n g r e l a t i v e l y simple b i o l o g i c a l endpoints (Underbrink et a l , 1976; L l o y d et a l , 1975) was not evident f o r e i t h e r V79 or CHO c e l l s u r v i v a l . S i m i l a r l y , the RBE of megavoltage e l e c t r o n s versus 6 0 C o 7-rays d i d not appear to decrease a p p r e c i a b l y w i t h dose, as has been p r e v i o u s l y suggested by e x t r a p o l a t i o n s of h i g h dose and f r a c t i o n a t e d dose data obtained u s i n g a human tumour c e l l l i n e (Amols et a l , 1986). Rather, the RBE's measured f o r both V79 and CHO c e l l s i n the t h e r a p e u t i c dose range were only s l i g h t l y d i f f e r e n t from RBE's seen at higher doses f o r 55 kV p X-rays, and even smaller d i f f e r e n c e s were seen f o r 250 kV p X-rays or 11 MeV e l e c t r o n s . These r e s u l t s t h e r e f o r e c o n t r a d i c t the p r e d i c t i o n s made by the Theory of Dual R a d i a t i o n A c t i o n on the b a s i s of m i c r o d o s i m e t r i c a l measurements i n 1 nm diameter spheres. The disagreement of the c e l l s u r v i v a l data w i t h the p r e d i c t i o n s of the TDRA was not e n t i r e l y unexpected, s i n c e , as discussed i n S e c t i o n 7.1.3, other aspects of t h i s theory that have been t e s t e d w i t h mammalian c e l l s u r v i v a l have not been supported experimentally. Moreover, c e l l k i l l i n g i s a more complex phenomenon than mutation or i n d u c t i o n of chromosome a b e r r a t i o n s , so assays measuring these types of damage may support the TDRA even i f c e l l s u r v i v a l data do not. Indeed, i t has been suggested by many authors that c e l l k i l l i n g may not occur i n d i r e c t p r o p o r t i o n to the damage i n i t i a l l y imparted. Some propose that n o n - s p e c i f i c c e l l u l a r damage ( i . e . damage that may t e m p o r a r i l y or permanently modify c e l l u l a r metabolism) may c o n t r i b u t e to the e f f e c t s of s p e c i f i c DNA damage on c e l l k i l l i n g (Booz and Feinendegen, 1988). Others suggest that the shoulder i n 218 the mammalian c e l l s u r v i v a l curve i s a t t r i b u t a b l e to s a t u r a b l e r e p a i r mechanisms i n the c e l l (e.g. A l p e r , 1980, 1984; Goodhead, 1982, 1985; Radford et a l , 1988) or through the a c t i o n of l i n e a r and qu a d r a t i c modes of damage r e p a i r and/or m i s r e p a i r (Tobias et a l , 1980; C u r t i s , 1986). Repair mechanisms are known to p l a y an important r o l e i n c e l l u l a r r a d i a t i o n response, since a l a r g e p r o p o r t i o n of the energy imparted to a c e l l by, f o r example, X-rays, does not lead to l e t h a l i n j u r y (Goodhead and Brenner, 1983). In the sa t u r a b l e r e p a i r models of c e l l u l a r radioresponse, i n i t i a l r a d i a t i o n - i n d u c e d l e s i o n s are assumed to be produced l i n e a r l y w i t h dose, and r e p a i r mechanisms are assumed to act through a s a t u r a b l e enzymatic process t h a t competes w i t h damage-fixation mechanisms i n the c e l l (Goodhead, 1985). I f c e l l s u r v i v a l i s indeed a f f e c t e d by a n o n - l i n e a r m o d i f i c a t i o n of the p h y s i c a l damage imparted, models such as the TDRA would not adequately p r e d i c t c e l l s u r v i v a l response, r e g a r d l e s s of whether the i n i t i a l damage i s produced i n a l i n e a r or a l i n e a r - q u a d r a t i c f a s h i o n , whether or not other s u r v i v a l models provide a more c o n s i s t e n t f i t to the data would, however, r e q u i r e f u r t h e r a n a l y s i s . P h y s i c a l reasons that account f o r the observed d e v i a t i o n s of s u r v i v a l measurements from the p r e d i c t i o n s of the TDRA have a l s o been proposed. The s i z e of the " s e n s i t i v e s i t e " i n which c r i t i c a l damage can occur may, f o r inst a n c e , be smaller than the micrometer dimensions p o s t u l a t e d by the TDRA. Indeed, i t has been suggested, on the b a s i s of s u r v i v a l measurements w i t h u l t r a s o f t X-rays (Goodhead, 1982; Goodhead and Brenner, 1983) and high energy ions (Zaider and R o s s i , 1985), that l e t h a l l e s i o n s are produced by energy d e p o s i t i o n s i n regions w i t h nanometer dimensions. As mentioned i n S e c t i o n 7.1.3 and the Appendix ( S e c t i o n 10.2), the p o s s i b i l i t y of l e s i o n p r o d u c t i o n over smaller dimensions than are measurable w i t h microdosimetric techniques, along w i t h more c a r e f u l c o n s i d e r a t i o n of the geometry of the 219 s e n s i t i v e s i t e , has been taken i n t o account i n the Ge n e r a l i z e d Theory of Dual R a d i a t i o n A c t i o n (GTDRA) ( K e l l e r e r and R o s s i , 1978). With appropriate assumptions, the modified theory has been found to be c o n s i s t e n t even w i t h u l t r a s o f t X-ray r e s u l t s (Brenner and Zaider, 1984), which have been considered a major piece of evidence against the b a s i c concepts of dual r a d i a t i o n a c t i o n . However, because knowledge of the geometry of the s e n s i t i v e s i t e i s l i m i t e d , the GTDRA i s more d i f f i c u l t to v e r i f y q u a n t i t a t i v e l y than i s the TDRA. C e l l s u r v i v a l data are d i f f i c u l t to o b t a i n w i t h s u f f i c i e n t p r e c i s i o n to a l l o w the unequivocal acceptance of a p a r t i c u l a r model of r a d i a t i o n a c t i o n . This l i m i t a t i o n was i l l u s t r a t e d q u i t e profoundly by the d i f f i c u l t i e s encountered i n a c c u r a t e l y determining values of the quadratic component f o r l i n e a r - q u a d r a t i c f i t s to c e l l s u r v i v a l data, e s p e c i a l l y i n the low dose range. Since the constancy of f o r d i f f e r e n t r a d i a t i o n m o d a l i t i e s i s a p r e d i c t i o n of the TDRA and the GTDRA, the i n a b i l i t y to determine i t s value a c c u r a t e l y makes t h i s aspect of the theory d i f f i c u l t to t e s t . Indeed, a n a l y s i s of the low dose data alone would have i n d i c a t e d s i g n i f i c a n t d i f f e r e n c e s i n f5 between t e s t and reference r a d i a t i o n s f o r some of the low LET m o d a l i t i e s , yet a combination of the low dose w i t h the high dose data y i e l d e d reasonably s i m i l a r values of /3 f o r the m o d a l i t i e s t e s t e d . The s i m i l a r /3 values obtained f o r the combined low dose/high dose f i t s , as w e l l as evidence of a s l i g h t l y dose-dependent RBE, could be considered to support some p r e d i c t i o n s of the TDRA. Nevertheless, i t i s c l e a r t h a t i f the TDRA were a p p l i c a b l e to the experimental data, i t would be i n i t s g e n e r a l i z e d form (the GTDRA) . Since a r a t i o s f o r d i f f e r e n t r a d i a t i o n m o d a l i t i e s cannot be p r e d i c t e d s t r i c t l y from microdosimetric measurements i n the GTDRA, i t cannot be s a i d whether or not t h i s theory i s f u l l y supported by the low dose RBE's measured i n t h i s t h e s i s . 220 8.2.3 Sources of E r r o r P a r t i c u l a r to the RBE Experiments In a d d i t i o n to the sources of v a r i a t i o n discussed i n S e c t i o n 8.1.3, there were s e v e r a l other f a c t o r s that may have a f f e c t e d the r e s u l t s obtained f o r the RBE experiments. Included among these are e r r o r s r e l a t e d to dosimetry and to heterogeneity i n the c e l l p o p u l a t i o n . Of the e r r o r s a s s o c i a t e d w i t h the dosimetry, those a r i s i n g from uneven d i s t r i b u t i o n of dose i n the c e l l suspension were of g r e a t e s t concern because they were d i f f i c u l t to q u a n t i f y . The magnitude of these inhomogeneities had been estimated (Table V I I I ) , but, because i r r a d i a t i o n s were performed i n s t i r r e d suspensions, i t was assumed that a l l c e l l s would r e c e i v e the same dose. How w e l l t h i s assumption h e l d f o r the short i r r a d i a t i o n times used i n the low dose experiments i s d i f f i c u l t to assess. However, i t should be noted that the s h o r t e s t i r r a d i a t i o n times were f o r those m o d a l i t i e s ( 6 0Co 7-rays and 11 MeV e l e c t r o n s ) which experienced very l i t t l e v a r i a t i o n i n dose w i t h i n the volume of the c e l l suspension. In c o n t r a s t , i r r a d i a t i o n times f o r 55 kV p X-rays, which were attenuated the most by the c e l l suspension, were s i g n i f i c a n t l y l a r g e r . Furthermore, the v a r i a b i l i t y i n the experimental r e s u l t s obtained f o r 55 kV p X-rays was no greater than f o r any other modality. Increased v a r i a b i l i t y would have been expected i f a la r g e non-systematic e r r o r , such as uneven dose d i s t r i b u t i o n combined w i t h inadequate mixing of the c e l l suspension, had been present. Moreover, hi g h dose RBE values obtained f o r the 55 kV p X-rays were comparable to those found by other i n v e s t i g a t o r s (see Table V I I ) . The dosimetry f o r t h i s modality was therefore probably not s e v e r e l y a f f e c t e d by inadequate mixing of the c e l l suspension during the i r r a d i a t i o n s . Heterogeneity i n the c e l l p o p u l a t i o n was a l s o a source of concern i n the RBE experiments, e s p e c i a l l y since i t has been observed by other i n v e s t i g a t o r s that there may be s t r u c t u r e i n asynchronous c e l l s u r v i v a l 221 curves a t low doses (Skarsgard e t a l , 1987 and 1989). An asynchronous c e l l p o p u l a t i o n , which a c t u a l l y represents a mixture of subpopulations w i t h d i f f e r e n t r a d i o s e n s i t i v i t i e s , was used f o r a l l the experiments performed f o r t h i s t h e s i s . I t has been argued ( H a l l , 1975; G i l l e s p i e et a l , 1975a) tha t parameter estimates made w i t h such a p o p u l a t i o n would not be very meaningful i n the t e s t i n g of models of r a d i a t i o n a c t i o n . Asynchrony may indeed have been r e s p o n s i b l e f o r some of the problems encountered i n the determination of s u r v i v a l curve parameters. However, i t i s u n l i k e l y that the e f f e c t s of r e l a t i v e l y l a r g e changes i n RBE, as were p r e d i c t e d by the TDRA, would have been obscured by the heterogeneity i n an asynchronous c e l l p o p u l a t i o n to such a degree that the experimentally observed changes i n RBE wi t h dose would be much l e s s than expected. Indeed, when the p r e d i c t i o n s of the TDRA were a p p l i e d uniformly to expe r i m e n t a l l y determined a and /3 values obtained by G i l l e s p i e et a l (1975b) f o r V79 c e l l s i n d i f f e r e n t phases of the c e l l c y c l e , a s u p e r p o s i t i o n of these values to simulate an e x p o n e n t i a l l y growing c e l l p o p u l a t i o n y i e l d e d o v e r a l l a r a t i o s almost i d e n t i c a l to those i n i t i a l l y a p p l i e d . By c o n t r a s t , / J 1 / 2 r a t i o s could deviate by a f a i r amount from the value of 1.0, i n d i c a t i n g t h a t parameters dependent on h i g h dose e f f e c t s are more s e n s i t i v e to heterogeneity than are those that are dependent on low dose e f f e c t s , at l e a s t according to the circumstances p r e d i c t e d by the TDRA. This above a n a l y s i s d i d not, however, take i n t o c o n s i d e r a t i o n the p o s s i b i l i t y that RBE e f f e c t s c o u l d vary w i t h i n the c e l l c y c l e . Only d i r e c t measurements w i t h t i g h t l y synchronized c e l l p o p ulations could r e s o l v e t h i s i s s u e . 8.2.4 Recommendations f o r Further I n v e s t i g a t i o n s While the data c o l l e c t e d i n t h i s t h e s i s were not s u f f i c i e n t l y e xtensive to prove or disprove the concept of dual r a d i a t i o n a c t i o n , they d i d p rovide an i n d i c a t i o n of the v a l i d i t y of c u r r e n t l y accepted RBE values f o r c e l l s u r v i v a l i n response to low LET r a d i a t i o n s . Accepted RBE values p r e d i c t e d from "high dose" data appear to apply w i t h reasonable accuracy to the doses used i n t y p i c a l r a d i a t i o n treatment f r a c t i o n s , a t l e a s t f o r asynchronous c e l l p o p ulations. There was, however, some i n d i c a t i o n i n t h i s t h e s i s that low dose RBE's are not always p r e d i c t e d w i t h a h i g h degree of accuracy by e x t r a p o l a t i o n s from high dose data ( c f . Table X I I ) . Because r a d i a t i o n treatment causes some r e d i s t r i b u t i o n of c e l l s i n the c e l l c y c l e , the p r o l i f e r a t i n g p o r t i o n of a tumour c e l l p o p u l a t i o n does not n e c e s s a r i l y remain asynchronous once a treatment regimen has been i n i t i a t e d . Further experiments using synchronized c e l l s would the r e f o r e have to be performed i n order to address the i m p l i c a t i o n s f o r radiotherapy more completely. Although i t i s u n l i k e l y that e f f e c t s as l a r g e as those p r e d i c t e d by the TDRA would be seen f o r c e l l s i n any phase of the c e l l c y c l e , i t i s p o s s i b l e that the RBE's of low LET r a d i a t i o n s vary to some degree w i t h c e l l c y c l e phase. Such v a r i a t i o n would not be unexpected f o r e i t h e r a l e s i o n - i n t e r a c t i o n or a n o n - l i n e a r r e p a i r model of c e l l s u r v i v a l . C e l l - c y c l e dependent changes i n the conformation of nuc l e a r DNA, which i s considered to be the s e n s i t i v e t a r g e t f o r c e l l k i l l i n g , c ould a f f e c t the e f f i c i e n c y w i t h which l e t h a l l e s i o n s are e i t h e r produced or r e p a i r e d . In a d d i t i o n to experiments w i t h synchronized c e l l s , courses of a c t i o n that may help provide a b e t t e r understanding of the mechanisms of r a d i a t i o n a c t i o n would i n c l u d e s t u d i e s comparing RBE's f o r c e l l s u r v i v a l w i t h RBE's f o r other endpoints, e s p e c i a l l y since l a r g e changes i n RBE have been observed w i t h decreasing dose f o r some b i o l o g i c a l e f f e c t s . Endpoints that 223 c o u l d be examined include mutation i n d u c t i o n , chromosome a b e r r a t i o n s , and formation of DNA double-strand breaks. The l a t t e r endpoint i s of considerable i n t e r e s t , e s p e c i a l l y i n l i g h t of recent r e p o r t s t h a t the l e v e l of mammalian c e l l k i l l i n g r e f l e c t s the l e v e l of induced DNA double-strand breaks when both are measured i n the same dose range (Radford, 1986, 1988). Knowledge of the r e l a t i o n s h i p between double-strand break occurrence and c e l l s u r v i v a l may provide some i n s i g h t i n t o the a s s o c i a t i o n between the i n i t i a l r a d i a t i o n - i n d u c e d DNA damage and the u l t i m a t e s u r v i v a l p r o b a b i l i t y f o r a c e l l , e s p e c i a l l y i f the occurrence of double s t r a n d breaks under c o n d i t i o n s where r e p a i r i s i n h i b i t e d could be st u d i e d . 224 9. BIBLIOGRAPHY Almeida, S.P. and F u j i i , H. (1979). F o u r i e r transform d i f f e r e n c e s and averaged s i m i l a r i t i e s i n diatoms. Appl. Opt. 18, 1663-1667. A l p e r , T. (1979). Cellular Radiobiology (Cambridge U n i v e r s i t y Press, Cambridge). A l p e r , T. (1980). S u r v i v a l curve models. In: Radiation Biology in Cancer Research, e d i t e d by R.E. Meyn and H.R. Withers (Raven Press, New York), pp. 3-18. Al p e r , T. (1984). I m p l i c a t i o n s of r e p a i r models f o r LET e f f e c t s and other r a d i o b i o l o g i c a l phenomena. Br. J. Cancer. 49, Suppl. VI, 137-143. Amols, H.I., Lagueux, B., and Cagna, D. (1986). R a d i o b i o l o g i c a l e f f e c t i v e n e s s (RBE) of megavoltage X-ray and e l e c t r o n beams i n radiotherapy. Radiat. Res. 105, 58-67. Armelin, H.A. and Armelin, M.C.S. (1978). R e s t i n g - p r o l i f e r a t i v e t r a n s i t i o n i n c u l t u r e d mammalian c e l l s . Natl. Cancer Inst. Monogr. 48, 81-85. Barendsen, G.W. (1968). Responses of c u l t u r e d c e l l s , tumours, and normal t i s s u e s to r a d i a t i o n s of d i f f e r e n t l i n e a r energy t r a n s f e r . Curr. Top. Radiat. Res. 4, 295-356. Bedford, J.S. and Griggs, H.G. (1975). The e s t i m a t i o n of s u r v i v a l at low doses and the l i m i t s of r e s o l u t i o n of the s i n g l e c e l l p l a t i n g technique. In: Cell Survival after Low Doses of Radiation, e d i t e d by T. Al p e r (John Wiley and Sons, London), pp. 34-39. B e l l i , M., Cher u b i n i , R., F i n o t t o , S., Moschini, G., Sapora, 0., Simone, G., and Tabocchini, M.A. (1989). RBE-LET r e l a t i o n s h i p f o r the s u r v i v a l of V79 c e l l s i r r a d i a t e d w i t h low energy protons. Int. J. Radiat. Biol. 55, 93-104. Bevington, P.R. (1969). Data Reduction and Error Analysis for the Physical Sciences (McGraw-Hill Book Company, New York). B i r d , R.P. (1979). B i o p h y s i c a l s t u d i e s w i t h s p a t i a l l y c o r r e l a t e d ions. 3. C e l l s u r v i v a l s t u d i e s using diatomic deuterium. Radiat. Res. 78, 210-223. Boag, J.W. (1975). The s t a t i s t i c a l treatment of c e l l s u r v i v a l data. In: Cell Survival after Low Doses of Radiation, e d i t e d by T. Al p e r (John Wiley and Sons, London), pp. 40-53. Bond, V.P., Meinhold, C.B., and R o s s i , H.H. (1978). Low-dose RBE and Q f o r X-ray compared to gamma-ray r a d i a t i o n s . Health Physics 34, 433-438. Booz, J . and Feinendegen, L.E. (1988). A microdosimetric understanding of low-dose r a d i a t i o n e f f e c t s . Int. J. Radiat. Biol. 53, 13-21. Brenner, D.J. and Zaider, M. (1984). M o d i f i c a t i o n of the Theory of Dual R a d i a t i o n A c t i o n f o r attenuated f i e l d s . I I . A p p l i c a t i o n to the a n a l y s i s of s o f t X-ray r e s u l t s . Radiat. Res. 99, 492-501. Brosin g , J . (1983). Measurements of Cell Survival at Low Doses of Radiation. Ph.D. Thesis, The U n i v e r s i t y of B r i t i s h Columbia, Vancouver, Canada. Carlson, F.P. and Lee, C.K. (1978). An adaptive coherent o p t i c a l processor f o r c e l l r e c o g n i t i o n and counting. IEEE Trans. Biomed. Eng. BME-25, 361-367. Casasent, D. (1985). Coherent o p t i c a l p a t t e r n r e c o g n i t i o n : a review. Opt. Eng. 24, 26-32. Chadwick, K.H. and Leenhouts, H.P. (1973). A molecular theory of c e l l s u r v i v a l . Phys. Med. Biol. 18, 78-87. Chapman, J.D. (1980). B i o p h y s i c a l models of mammalian c e l l i n a c t i v a t i o n by r a d i a t i o n . In: Radiation Biology in Cancer Research, e d i t e d by R.E. Meyn and H.R. Withers (Raven Press, New Y o r k ) , pp. 21-32. Cox, R., Thacker, J . , and Goodhead, D.T. (1977). I n a c t i v a t i o n and mutation of c u l t u r e d mammalian c e l l s by aluminium c h a r a c t e r i s t i c u l t r a s o f t X-rays. I I . Dose-response of Chinese hamster and human d i p l o i d c e l l s to aluminium X-rays and r a d i a t i o n s of d i f f e r e n t LET. Int. J. Radiat. Biol. 31, 561-576. C u r t i s , S.B. (1986). L e t h a l and p o t e n t i a l l y l e t h a l l e s i o n s induced by r a d i a t i o n - a u n i f i e d r e p a i r model. Radiat. Res. 106, 252-270. Dodd, J . and McCrone, W.C. (1975). A S c h l i e r e n eyepiece. The Microscope 23, 89-92. Denekamp, J . (1982). Cell Kinetics and Cancer Therapy (Charles C. Thomas, P u b l i s h e r , S p r i n g f i e l d , I l l i n o i s ) . E l k i n d , M.M., Han, A., and V o l z , K.W. (1963). R a d i a t i o n response of mammalian c e l l s grown i n c u l t u r e . IV. Dose dependence of d i v i s i o n delay and p o s t i r r a d i a t i o n growth of s u r v i v i n g and n o n s u r v i v i n g Chinese hamster c e l l s . J. Nat. Cancer Inst. 30, 705-721. E l l e t t , W.H. and Braby, L.A. (1972). The microdosimetry of 250 kVp and 65 kVp X-rays, 6 0Co gamma rays, and t r i t i u m beta p a r t i c l e s . Radiat. Res. 51, 229-243. Faddegon, B. (1983). The Effect of Misonidazole on Cell Survival at Low Doses of Radiation. M.Sc. Thesis, The U n i v e r s i t y of B r i t i s h Columbia, Vancouver, Canada. Fo l k a r d , M., P r i s e , K.M., Vojnovic, B., Davies, S., Roper, M.J., and Michael, B.D. (1989). The i r r a d i a t i o n of V79 mammalian c e l l s by protons w i t h energies below 2 MeV. Int. J. Radiat. Biol. 56, 221-237. 226 Franko, A.J. and Sutherland, R.M. (1979). Oxygen d i f f u s i o n d i s t a n c e and development of n e c r o s i s i n m u l t i c e l l spheroids. Radiat. Res. 79, 439-453. Freshney, R.I. (1987). Culture of Animal Cells, 2 n d e d i t i o n (Alan R. L i s s , Inc., New York). F r i c k e , H. and Morse, S. (1929). The a c t i o n s of X-rays on f e r r o u s sulphate s o l u t i o n s . Phil. Mag. 7, 129. G i l l e s p i e , C.J., Chapman, J.D., Reuvers, A.P., and Dugle, D.L. (1975a). S u r v i v a l of X - i r r a d i a t e d hamster c e l l s : a n a l y s i s i n terms of the Chadwick-Leenhouts model. In: Cell Survival after Low Doses of Radiation, e d i t e d by T. Alper (John Wiley and Sons, London), pp. 25-33. G i l l e s p i e , C.J., Chapman, J.D., Reuvers, A.P., and Dugle, D.L. (1975b). The i n a c t i v a t i o n of Chinese hamster c e l l s by X-rays: Synchronized and exponential c e l l p o p ulations. Radiat. Res. 64, 353-364. Goodhead, D.T. (1977). I n a c t i v a t i o n and mutation of c u l t u r e d mammalian c e l l s by aluminium c h a r a c t e r i s t i c u l t r a s o f t X-rays. I I I . I m p l i c a t i o n s f o r the theory of dual r a d i a t i o n a c t i o n . Int. J. Radiat. Biol. 32, 43-70. Goodhead, D.T. (1980). Models of r a d i a t i o n i n a c t i v a t i o n and mutagenesis. In: Radiation Biology in Cancer Research, e d i t e d by R.E. Meyn and H.R. Withers (Raven Press, New Y o r k ) , pp. 231-247. Goodhead, D.T. (1982). An assessment of the r o l e of microdosimetry i n r a d i o b i o l o g y . Radiat. Res. 91, 45-76. Goodhead, D.T. (1985). Saturable r e p a i r models of r a d i a t i o n a c t i o n i n mammalian c e l l s . Radiat. Res. 104, S58-S67. Goodhead, D.T. (1987). R e l a t i o n s h i p of microdosimetric techniques to a p p l i c a t i o n s i n b i o l o g i c a l systems. In: The Dosimetry of Ionizing Radiation, V o l . I I , e d i t e d by K.R. Kase, B.E. Bjarngard, and F.H. A t t i x (Academic Press, Inc., New Y o r k ) , pp. 1-89. Goodhead, D.T. and Brenner, D.J. (1983). E s t i m a t i o n of a s i n g l e property of low LET r a d i a t i o n s which c o r r e l a t e s w i t h b i o l o g i c a l e f f e c t i v e n e s s . Phys. Med. Biol. 28, 485-492. Goodman, J.W. (1968). Introduction to Fourier Optics (McGraw-Hill Book Company, New York). Grote, S.J., J o s h i , G.P., R e v e l l , S.H., and Shaw, C.A. (1981a). A method f o r the s c r u t i n y of l i v e mammalian c e l l s i n c u l t u r e and f o r the measurement of t h e i r p r o l i f e r a t i v e a b i l i t y a f t e r X - i r r a d i a t i o n . I n t . J. Radiat. Biol. 39, 377-394. Grote, S.J., J o s h i , G.P., R e v e l l , S.H., and Shaw, C.A. (1981b). Observations of r a d i a t i o n - i n d u c e d chromosome fragment l o s s i n l i v e mammalian c e l l s i n c u l t u r e , and i t s e f f e c t on colony-forming a b i l i t y . Int. J. Radiat. Biol. 39, 395-408. 227 H a l l , E.J. (1975). B i o l o g i c a l problems i n the measurement of s u r v i v a l at low doses. In: Cell Survival after Low Doses of Radiation, e d i t e d by T. Alper (John Wiley and Sons, London), pp. 13-24. H a l l , E.J. (1988). Radiobiology for the Radiologist, 3 r d E d i t i o n (J.B. L i p p i n c o t t Company, P h i l a d e l p h i a ) . H a l l , E.J., Kovak, J.K., K e l l e r e r , A.M., R o s s i , H.H., Marino, S., and Goodman, L.J . (1975). RBE as a f u n c t i o n of neutron energy. I. Experimental observations. Radiat. Res. 64, 245-255. Hecht, E. (1987). Optics, 2 n d E d i t i o n (Addison-Wesley P u b l i s h i n g Company, Reading, Massachusetts). Hewitt, H.B. and Wilson, C.W. (1959). A s u r v i v a l curve f o r mammalian c e l l s i r r a d i a t e d in vivo. Nature 183, 1060-1061. Heye, R.R., K i e b l e r , E.W., and Armzen, R.J. (1982). M u l t i p l e x e d time lapse photomicrography of c u l t u r e d c e l l s . J. Microscopy 125, 41-50. Hiraoka, Y., Sedat, J.W., and Agard, D.A. (1987). The use of a charge-coupled device f o r q u a n t i t a t i v e o p t i c a l microscopy of b i o l o g i c a l s t r u c t u r e s . Science 238, 36-41. Hoffman, R. and Gross, L. (1975). Modulation c o n t r a s t microscope. Appl. Opt. 14, 1169-1176. Hoshi, M., Antoku, S., Nakamura, N. , R u s s e l l , W.J., M i l l e r , R.C., Sawada, S., Mizuno, M., and N i s h i o , S. (1988). S o f t X-ray dosimetry and RBE f o r s u r v i v a l of Chinese hamster V79 c e l l s . Int. J. Radiat. Biol. 54, 577-591. H u t z l e r , P.J.S. (1977). S p a t i a l frequency f i l t e r i n g and i t s a p p l i c a t i o n to microscopy. Appl. Opt. 16, 2264-2272. ICRP (1955) . Recommendations of the I n t e r n a t i o n a l Commission on R a d i o l o g i c a l P r o t e c t i o n , Brit. J. Radiol., Suppl. No. 6, London. ICRU (1969). Radiation Dosimetry: X Rays and Gamma Rays with Maximum Photon Energies Between 0.6 and 50 Mev, ICRU Report 14 ( I n t e r n a t i o n a l Commission on R a d i a t i o n U n i t s and Measurements, Bethesda, Maryland). ICRU (1970a). Linear Energy Transfer, ICRU Report 16 ( I n t e r n a t i o n a l Commission on R a d i a t i o n U n i t s and Measurements, Bethesda, Maryland). ICRU (1970b). Radiation Dosimetry: X Rays Generated at P o t e n t i a l s of 5 to 150 kV, ICRU Report 17 ( I n t e r n a t i o n a l Commission on R a d i a t i o n U n i t s and Measurements, Bethesda, Maryland). ICRU (1980). Radiation Quantities and Units, ICRU Report 33 ( I n t e r n a t i o n a l Commission on R a d i a t i o n U n i t s and Measurements, Bethesda, Maryland). ICRU (1984). Radiation Dosimetry: Electron Beams with Energies Between 1 and 50 MeV, ICRU Report 35 ( I n t e r n a t i o n a l Commission on R a d i a t i o n U n i t s and Measurements, Bethesda, Maryland). 228 ICRU (1986). The Quality Factor in Radiation Protection, ICRU Report 40 ( I n t e r n a t i o n a l Commission on R a d i a t i o n U n i t s and Measurements, Bethesda, Maryland). Inoue, S. (1986). Video Microscopy (Plenum Press, New York). J a g g i , B. and P a l c i c , B. (1985). The design and development of an o p t i c a l scanner f o r c e l l b i o l o g y . IEEE Proc. Eng. Med. Biol. 2, 980-985. J a g g i , B., Poon, S.S.S., and P a l c i c , B. (1986). Implementation and e v a l u a t i o n of the DMIPS C e l l Analyzer. IEEE Proc. Eng. Med. Biol. 3, 906-911. James, M. (1985). C l a s s i f i c a t i o n Algorithms ( W i l l i a m C o l l i n s Sons and Co. L t d . , London). J e n n r i c h , R. and Sampson, P. (1983). Stepwise d i s c r i m i n a n t a n a l y s i s . In: BMDP S t a t i s t i c a l Software ( r e v i s e d p r i n t i n g ) , e d i t e d by W.J. Dixon ( U n i v e r s i t y of C a l i f o r n i a Press, B e r k e l e y ) , pp. 519-537. Johns, H.E. and Cunningham, J.R. (1983). The Physics of Radiology, 4 t h e d i t i o n (Charles C. Thomas, P u b l i s h e r , S p r i n g f i e l d , I l l i n o i s ) . Kallman, R.F. (1984). Automated autoradiographic a n a l y s i s of tumor c e l l c o l o n i e s in vitro. In: B i o l o g i c a l Dosimetry, e d i t e d by W.G. E i s e r t and M.L. Mendelsohn (Springer V e r l a g , B e r l i n ) , pp. 255-263. Kallman, R.F., Combs, C.A., Franko, A.J., Furlong, B.M., K e l l e y , S.D., Kemper, H.L., M i l l e r , R.G., Rapacchietta, D., Schoenfeld, D., and Takahashi, M. (1980). Evidence f o r the recruitment of n o n c y c l i n g clonogenic tumor c e l l s . In: Radiation Biology in Cancer Research, e d i t e d by R.E. Meyn and H.R. Withers (Raven Press, New Y o r k ) , pp. 397-414. K e l l e r e r , A.M. (1985). Fundamentals of microdosimetry. In: The Dosimetry of Ionizing Radiation, V o l . I , e d i t e d K.R. Kase, B.E. Bjarngard, and F.H. A t t i x (Academic Press, Inc., New York), pp. 77-162. K e l l e r e r , A.M. and Chmelevsky, D. (1975). C r i t e r i a f o r the a p p l i c a b i l i t y of LET. Radiat. Res. 63, 226-234. K e l l e r e r , A.M., H a l l , E.J., R o s s i , H.H., and Teedla, P. (1976). RBE as a f u n c t i o n of neutron energy. I I . S t a t i s t i c a l a n a l y s i s . Radiat. Res. 65, 172-186. K e l l e r e r , A.M., Lam, Y.P., and R o s s i , H.H. (1979). B i o p h y s i c a l s t u d i e s w i t h s p a t i a l l y c o r r e l a t e d ions. 4. A n a l y s i s of c e l l s u r v i v a l data f o r diatomic deuterium. Radiat. Res. 83, 511-528. K e l l e r e r , A.M. and R o s s i , H.H. (1971). RBE and the primary mechanism of r a d i a t i o n a c t i o n . Radiat. Res. 47, 15-34. K e l l e r e r , A.M. and R o s s i , H.H. (1972). The theory of dual r a d i a t i o n a c t i o n . Curr. Top. Radiat. Res. 8, 85-158. 229 K e l l e r e r , A.M. and R o s s i , H.H. (1978). A g e n e r a l i z e d f o r m u l a t i o n of dual r a d i a t i o n a c t i o n . Radiat. Res. 75, 471-488. K r a f t , G. and Kraft-Weyrather, W. (1987). B i o p h y s i c a l aspects of t r a c k s t r u c t u r e . Proceedings of the 8th International Congress of Radiation Research, V o l . 2, pp. 29-34. Lea, D.E. (1955). Actions of Radiations on Living Cells, 2 n d E d i t i o n (Cambridge U n i v e r s i t y Press, Cambridge). L l o y d , D.C, P u r r o t t , R.J., Dolphin, G.W. , B olton, D. , Edwards, A.A. , and Corp, M.J. (1975). The r e l a t i o n s h i p between chromosome a b e r r a t i o n s and low LET r a d i a t i o n dose to human lymphocytes. Int. J. Radiat. Biol. 28, 75-90. Malone, J.F., P o r t e r , D., and Hendry, J.H. (1974). The r.b.e. of 6 0 C o gamma rays w i t h respect to 300 kVp X-rays f o r the s u r v i v a l of HeLa S-3 and CHO c e l l s , i r r a d i a t e d i n d i f f e r e n t s t a t e s of p r o l i f e r a t i o n . Int. J. Radiat. Biol. 26, 355-362. Marin, G. and Bender, M.A. (1963). A comparison of mammalian c e l l - k i l l i n g by i n c o r p o r a t e d H 3-thymidine and H 3 - u r i d i n e . Int. J. Radiat. Biol. 7, 235-244. McCrone, W.C. (1975). Determination of n D, n p, and n c by d i s p e r s i o n s t a i n i n g . The Microscope 23, 213-220. M i l l a r , B.C., F i e l d e n , E.M., and M i l l a r , J.L. (1978). I n t e r p r e t a t i o n of s u r v i v a l - c u r v e data f o r Chinese hamster c e l l s , l i n e V-79 u s i n g the m u l t i - t a r g e t , m u l t i - t a r g e t w i t h i n i t i a l slope, and a, 8 equations. Int. J. Radiat. Biol. 33, 599-603. Munro, T.R. (1970). The r e l a t i v e r a d i o s e n s i t i v i t y of the nucleus and cytoplasm of Chinese hamster f i b r o b l a s t s . Radiat. Res. 42, 451-470. Nias, A.H.W., G i l b e r t , C.W., L a j t h a , L.G. , and Lange, C S . (1965). Cl o n e - s i z e a n a l y s i s i n the study of c e l l growth f o l l o w i n g s i n g l e or during continuous i r r a d i a t i o n . Int. J. Radiat. Biol. 9, 275-290. Ojeda-Castaneda, J . (1978). Foucault, wire, and phase modulation t e s t s . In: Optical Shop Testing, e d i t e d by D. Malacara (John Wiley and Sons, New Yo r k ) , pp. 231-281. P a l c i c , B., Brosing, J.W., Lam, G.Y.K., and Skarsgard, L.D. (1985). The requirement f o r s u r v i v a l measurements at low doses. In: Proceedings of the Berkeley Conference in Honor of Jerzy Neyman and Jack Kiefer, Volume I , e d i t e d by L.M. Le Cam and R.A. Olshen (Wadsworth, Inc., B e r k e l e y ) , pp. 331-341. P a l c i c , B., Faddegon, B., J a g g i , B., and Skarsgard, L.D. (1983). Automated low dose assay system f o r s u r v i v a l measurements of mammalian c e l l s in vitro. J. Tiss. Cult. Meth. 8, 103-107. P a l c i c , B., Faddegon, B., and Skarsgard, L.D. (1984). The e f f e c t of misonidazole as a hypoxic r a d i o s e n s i t i z e r at low dose. Radiat. Res. 100, 340-347. 230 P a l c i c , B. and J a g g i , B. (1986). The use of s o l i d s t a t e image sensor technology to detect and c h a r a c t e r i z e l i v e mammalian c e l l s growing i n t i s s u e c u l t u r e . I n t . J. Radiat. Biol. 50, 345-352. P a l c i c , B. and J a g g i , B. (1990). Image cytometry system f o r morphometric measurements of l i v e c e l l s . In: Bio instrument at ion: Developments and Applications, e d i t e d by D.L. Wise (Butterworth P u b l i s h e r s , Stoneham, MA), pp. 923-991. Parker, L., Skarsgard, L.D., and Emmerson, P.T. (1969). S e n s i t i z a t i o n of anoxic mammalian c e l l s to X-rays by triacetoneamine-N-oxyl. S u r v i v a l and t o x i c i t y s t u d i e s . Radiat. Res. 38, 493-500. Pernick, B., Kopp, R.E., L i s a , J . , Mendelsohn, J . , Stone, H., and Wohlers, R. (1978). Screening of c e r v i c a l c y t o l o g i c a l samples u s i n g coherent o p t i c a l processing. Part 1. Appl. Opt. 17, 21-34. Ploem, J.S., Verwoerd, N., Bonnet, J . , and Koper, G. (1979). An automated microscope f o r q u a n t i t a t i v e cytology combining t e l e v i s i o n image a n a l y s i s and stage scanning microphotometry. J. Histochem. Cytochem. 27, 136-143. P l u t a , M. (1988). Advanced Light Microscopy, V o l . 1 ( E l s e v i e r Science P u b l i s h i n g Company, New York). Poon, S.S.S., J a g g i , B., and P a l c i c , B. (1987). C e l l r e c o g n i t i o n algorithms f o r the C e l l Analyzer. IEEE Proc. Eng. Med. Biol. 4, 1454-1456. Poon, S.S.S., J a g g i , B., Spadinger, I . , and P a l c i c , B. (1989). Focussing methods used i n the C e l l Analyzer. IEEE Proc. Eng. Med. Biol. 11, 754-756. Puck, T.T. and Marcus, P.I. (1955). A r a p i d method f o r v i a b l e c e l l t i t r a t i o n and clone production w i t h HeLa c e l l s i n t i s s u e c u l t u r e . Proc. Nat. Acad. Sc. 41, 432. Puck, T.T. and Marcus, P.I. (1956). A c t i o n of X-rays on mammalian c e l l s . J. Exp. Med. 103, 653-666. Puck, T.T., Marcus, P.I., and C i e c i u r a , S.J. (1956). C l o n a l growth of mammalian c e l l s i n v i t r o . J. Exp. Med. 103, 273-283. Puck, T.T., Morkovin, D., Marcus, P.I., and C i e c i u r a , S.J. (1957). A c t i o n of X-rays on mammalian c e l l s I I - S u r v i v a l curves of c e l l s from normal human t i s s u e s . J. Exp. Med. 106, 485-500. Q u i n t i l i a n i , M. (1979). M o d i f i c a t i o n of r a d i a t i o n s e n s i t i v i t y : the oxygen e f f e c t . Int. J. Radiation Oncology Biol. Phys. 5, 1069-1076. Radford, I.R. (1986). Evidence f o r a general r e l a t i o n s h i p between the induced l e v e l of DNA double-strand breakage and c e l l - k i l l i n g a f t e r X - i r r a d i a t i o n of mammalian c e l l s . Int. J. Radiat. Biol. 49, 611-620. 231 Radford, I.R. (1988). The dose-response f o r low-LET r a d i a t i o n - i n d u c e d DNA double-strand breakage: methods of measurement and i m p l i c a t i o n s f o r r a d i a t i o n a c t i o n models. I n t . J. Radiat. Biol. 54, 1-11. Radford, I.R., Hodgson, G.S., and Matthews, J.P. (1988). C r i t i c a l DNA tar g e t s i z e model of i o n i z i n g r a d i a t i o n - i n d u c e d c e l l death. I n t . J . Radiat. Biol. 54, 63-79. Ratkowsky, D.A. (1983). Nonlinear Regression Modeling (Marcel Dekker, Inc. , New York) . Reddy, N.M.S., Stevenson, A.F.G., and Lange, C S . (1989). T r y p s i n i z a t i o n and the r a d i o s e n s i t i v i t y of m i t o t i c and l o g phase Chinese hamster V79 c e l l s to 250 kVp X-rays. Int. J. Radiat. Biol. 55, 105-117. Rochow, T.G. and Rochow, E.G. (1978). An Introduction to Microscopy by Means of Light, Electrons, X-rays, or Ultrasound (Plenum Press, New York). R o n d a n e l l i , E.G., Trenta, A., M a g l i u l o , E., Va n n i n i , V. and Gerna, G. (1966). Giant c e l l s produced by m i t o t i c r e v e r s i o n i n prophase, metaphase, anaphase a f t e r low doses of r a d i a t i o n . Exp. Cell Res. 41, 1-11. Roots, R. and Okada, S. (1975). E s t i m a t i o n of l i f e times and d i f f u s i o n d i s t a n c e s of r a d i c a l s i n v o l v e d i n X-ray-induced DNA st r a n d breaks or k i l l i n g of mammalian c e l l s . Radiat. Res. 64, 306-320. R o s s i , H.H. (1959). S p e c i f i c a t i o n of r a d i a t i o n q u a l i t y . Radiat. Res. 10, 522-531. R o s s i , H.H. (1977). A proposal f o r the r e v i s i o n of the q u a l i t y f a c t o r . Radiat. Environ. Biophys. 14, 275-283. R o s s i , H.H. (1978). R a d i a t i o n q u a l i t y and q u a l i t y f a c t o r . In: Sixth Symposium on Microdosimetry, V o l . I I , e d i t e d by J . Booz and H.G. Ebert (Harwood Academic P u b l i s h e r s L t d . , London), pp. 1087-1097. R o s s i , H.H. (1986). R a d i a t i o n q u a l i t y . Radiat. Res. 107, 1-10. Schmid, E., Rimpl, G., and Bauchinger, M. (1974). Dose-response r e l a t i o n of chromosome abe r r a t i o n s i n human lymphocytes a f t e r in vitro i r r a d i a t i o n w i t h 3 MeV e l e c t r o n s . Radiat. Res. 57, 228-238. S i n c l a i r , W.K. (1962). The r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s of 22-Mevp X-rays, cobalt-60 gamma rays, and 200 Kvcp X-rays. V I I . Summary of stu d i e s f o r f i v e c r i t e r i a of e f f e c t . Radiat. Res. 16, 394-398. S i n c l a i r , W.K. (1964). X-ray induced h e r i t a b l e damage (small-colony formation) i n c u l t u r e d mammalian c e l l s . Radiat. Res. 21, 584-611. S i n c l a i r , W.K. and Kohn, H.I. (1964). The r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s of high energy photons and e l e c t r o n s . Radiology 82, 800-815. 232 Skarsgard, L.D., H a r r i s o n , I . , and Durand, R.E. (1987). C e l l s u r v i v a l at low dose: evidence f o r two components. A b s t r a c t C32-2P, Proceedings of the 8th International Congress of Radiation .Research, V o l . 1, p. 129. Skarsgard, L.D., Kihlman, B.A., Parker, L. , Pujara, CM., and Richardson, S. (1967). S u r v i v a l , chromosome a b n o r m a l i t i e s , and recovery i n heavy-ion- and x - i r r a d i a t e d mammalian c e l l s . Radiat. Res. Suppl. 7, 208-221. Skarsgard, L.D., Wilson, D.J., and Durand, R.E. (1989). R a d i a t i o n s u r v i v a l response at low dose i n V79-171 c e l l s . A b s t r a c t Co-7, 37th Annual Meeting of the Radiation Research Society. Spadinger, I . , Poon, S.S.S., and P a l c i c , B. (1989). Automated d e t e c t i o n and r e c o g n i t i o n of l i v e c e l l s i n t i s s u e c u l t u r e u s i n g image cytometry. Cytometry 10, 375-381. Spadinger, I . , Poon, S.S.S., and P a l c i c , B. (1990). E f f e c t of focus on c e l l d e t e c t i o n and r e c o g n i t i o n by the C e l l Analyzer. Cytometry 11, 460-467. S u i t , H.D. and Walker, A.M. (1989). P r e d i c t o r s of r a d i a t i o n response i n use today: C r i t e r i a f o r new assays and methods of v e r i f i c a t i o n . In: Prediction of Tumor Treatment Response, e d i t e d by J.D. Chapman, L.J. Peters, and H.R. Withers (Pergamon Press, New Y o r k ) , pp. 3-19. Sutherland, R.M., McCredie, J.A., and Inch, W.R. (1971). Growth of m u l t i c e l l spheroids i n t i s s u e c u l t u r e as a model of nodular carcinomas. J. Nat. Cancer Inst. 46, 113-120. Tanaka, N., Ikeda, H., Ueno, T., Watanabe, S., Imasato, Y., and Kashida, R. (1977). Fundamental study on automatic c y t o - s c r e e n i n g f o r u t e r i n e cancer. I I I . New system of automated apparatus (CYBEST) u t i l i z i n g the p a t t e r n r e c o g n i t i o n method. Acta Cytol. 21, 85-89. Thompson, B.J. (1978). O p t i c a l transforms and coherent p r o c e s s i n g systems - w i t h i n s i g h t s from c r y s t a l l o g r a p h y . In: Optical Data Processing, e d i t e d by D. Casasent (Springer V e r l a g , B e r l i n ) , pp. 17-52. Thurston, G., J a g g i , B., and P a l c i c , B. (1988). Measurement of c e l l m o t i l i t y and morphology w i t h an automated microscope system. Cytometry 9, 411-417. T i l l , J.E. and McCulloch, E.A. (1961). A d i r e c t measurement of the r a d i a t i o n s e n s i t i v i t y of normal mouse bone marrow c e l l s . Radiat. Res. 14, 213-222. Tobias, CA. , B l a k e l y , E.A., Ngo, F.Q.H., and Yang, T.C.H. (1980). The r e p a i r - m i s r e p a i r model of c e l l s u r v i v a l . In: Radiation Biology in Cancer Research, e d i t e d by R.E. Meyn and H.R. Withers (Raven Press, New Y o r k ) , pp. 195-230. Tolmach, A.P., M i t z , A.R., and VonRump, S.L. (1978). Computer-assisted a n a l y s i s of time lapse cinemicrographs of c u l t u r e d c e l l s . Comp. Biomed. Res. 11, 363-379. 233 Tucker, J.H. (1979). An image a n a l y s i s system f o r c e r v i c a l c y t o l o g y automation u s i n g nuclear DNA content. J. Histochem. Cytochem. 27, 613-620. Underbrink, A.G., K e l l e r e r , A.M., M i l l s , R.E. and Sparrow, A.H. (1976). Comparison of X-ray and gamma-ray dose-response curves f o r pink somatic mutations i n Tradescantia clone 02. Radiat. Environ. Biophys. 13, 295-303. V i d a l , B., S c h l u t e r , G., and Moore, G.W. (1973). C e l l nucleus p a t t e r n r e c o g n i t i o n : Influence of s t a i n i n g . Acta. Cytol. 17, 510-515. V i r s i k , R.P., Harder, D., and Hansmann, I. (1977). The RBE of 30 kV X-rays f o r the i n d u c t i o n of d i c e n t r i c chromosomes i n human lymphocytes. Radiat. Environ. Biophys. 14, 109-121. W i l l i a m s , P.C. and Hendry, J.H. (1978). The RBE of megavoltage photon and e l e c t r o n beams. Br. J. Radiol. 51, 220. Withers, H.R. and E l k i n d , M.M. (1969). R a d i o s e n s i t i v i t y and f r a c t i o n a t i o n response of c r y p t c e l l s of mouse jejunum. Radiat. Res. 38, 598-613. Young, M.E.J. (1983). Radiological Physics, 3 r d e d i t i o n (H.K. Lewis and Co. L t d . , London). Zaider, M., Brenner, D.J., Hanson, K., andMinerbo, G.N. (1982). An a l g o r i t h m f o r determining the p r o x i m i t y d i s t r i b u t i o n from dose-averaged l i n e a l energies. Radiat. Res. 91, 95-103. Zaider, M. and Brenner, D.J. (1985). On the microdosimetric d e f i n i t i o n of q u a l i t y f a c t o r s . Radiat. Res. 103, 302-316. Zaider, M. and R o s s i , H.H. (1985). Dual r a d i a t i o n a c t i o n and the i n i t i a l slope of s u r v i v a l curves. Radiat. Res. 104, S68-S76. Zaider, M. and R o s s i , H.H. (1988). On the a p p l i c a t i o n of microdosimetry to r a d i o b i o l o g y . Radiat. Res. 113, 15-24. Z e i t z , L., Kim, S.H., Kim, J.H., and Detko, J.F. (1977). Determination of r e l a t i v e b i o l o g i c a l e f f e c t i v e n e s s (RBE) of s o f t X-rays. Radiat. Res. 70, 552-563. Zeman, E.M. and Bedford, J.S. (1984). Changes i n e a r l y and l a t e e f f e c t s w i t h d o s e - p e r - f r a c t i o n : alpha, beta, r e d i s t r i b u t i o n and r e p a i r . I n t . J. Radiation Oncology Biol. Phys. 10, 1039-1047. Zernike, F. (1934). D i f f r a c t i o n theory of knife-edge t e s t and i t s improved form, the phase c o n t r a s t . Hon. Not. R. Astron. Soc. 94, 371. Zernike, F. (1958). The wave theory of microscopic image formation. In: Concepts of Classical Optics, e d i t e d by J . Strong (Freeman, San F r a n c i s c o ) , pp. 525-536. Z i r k l e , R.E., Marchbank, D.F., and Kuck, K.D. (1952). Exponential and sigmoid s u r v i v a l curves r e s u l t i n g from alpha and X - i r r a d i a t i o n of Aspergillus spores. J. Cell Comp. Physiol. 39, Suppl. 1, 75-85. 234 10. APPENDIX 10.1 Disc r i m i n a n t Function A n a l y s i s * D i s c r i m i n a n t f u n c t i o n s used to c l a s s i f y o b jects i n t o groups on the b a s i s of a set of feature values are g e n e r a l l y based on Bayes' C l a s s i f i c a t i o n Rule. Bayes' Rule s t a t e s : assign the object to group G. i f : P(G.|x) > P(G |x) for all j / i (10.1) where x i s the feature v e c t o r f o r the object. The a p p l i c a t i o n of t h i s r u l e can be s i m p l i f i e d using Bayes' Theorem: P(x|G k) P(G k) P ( G j x ) = (10.2) S P(x|G k) P(G k) . k C o n d i t i o n a l p r o b a b i l i t i e s of the form P(x|G k) are e a s i e r to determine from po p u l a t i o n sampling than are those of the form P(G |x). Nevertheless, e s t i m a t i o n of P(x|G k) s t i l l r e quires the c o l l e c t i o n and a n a l y s i s of large volumes of data, unless c e r t a i n assumptions are made about the nature of the feature d i s t r i b u t i o n s . Thus, features are g e n e r a l l y assumed to be normally d i s t r i b u t e d f o r each group, so that c o n d i t i o n a l p r o b a b i l i t i e s can be c a l c u l a t e d according to the m u l t i v a r i a t e normal d i s t r i b u t i o n : P(x|G k) = [(2 7r) n / 2|E kr / 2]- 1exp[-(x- M k)'S k- 1(x-M k)/2] (10.3) * References used i n the p r e p a r a t i o n of S e c t i o n 10.1 were James (1985) and J e n n r i c h and Sampson (1983) where n i s the number of fe a t u r e s , pfc i s the ve c t o r of feat u r e means f o r group k, and ^ i s the feature covariance matrix f o r group k. A f t e r s u b s t i t u t i o n , c a n c e l l i n g of common terms, and t a k i n g the loga r i t h m to el i m i n a t e exponential terms, Bayes' Rule f o r the normal case becomes: assign the object to group Gi if: d ^ x ) - I n f P ^ ) ] < d.(x) - ln[P(G.)] f o r all j / i (10.4) where l n | S k | + ( X - p k ) ' S k - 1 ( x - M k ) < y x ) . (10.5) This v e r s i o n of Bayes' Rule i s r e f e r r e d to as the quadratic d i s c r i m i n a n t f u n c t i o n , since expansion of the matrices y i e l d s a set of quadratic equations. Surfaces d i v i d i n g groups i n the feature space are there f o r e a l s o quadratic. This i s g r a p h i c a l l y i l l u s t r a t e d i n Figure 61a f o r the 2-group, 2-feature case. I f a d d i t i o n a l assumptions are made about the nature of the feature d i s t r i b u t i o n s , the c l a s s i f i c a t i o n r u l e can be f u r t h e r s i m p l i f i e d . In p a r t i c u l a r , i f the covariance matrices are equal f o r a l l the groups, Bayes' Rule can be expressed as: assign the object to group Gi if: f..(x) + in[P(G.)] > f . ( x ) + In[P(G.)] for all j / i (10.6) where f k ( X ) = v s - 1 * • (^'ir\)/2. (10.7) 2 3 6 U 3 (a) / ^ ^ ^ i ^ X group B group A /quadratic / decision boundary feature 1 feature 1 Figure 61. D i v i s i o n of a 2-dimensional feature space by a discriminant function. (a) quadratic discriminant function, (b) l i n e a r discriminant function (case of equal covariance matrices). E l l i p s e s represent equal-p r o b a b i l i t y contours of normally d i s t r i b u t e d , b i v a r i a t e feature data. 237 In t h i s case, E i s the common covariance matrix f o r a l l the groups, and l i n e a r surfaces d i v i d e the groups i n the feature space (see Figure 61b). Because i t i s more e a s i l y c a l c u l a t e d and a p p l i e d , the l i n e a r c l a s s i f i c a t i o n r u l e i s used more commonly i n p r a c t i c e than i s the quadratic r u l e , even when the covariance matrices f o r the d i f f e r e n t groups are not equal. In some instances, i t i s d e s i r a b l e to combine a fe a t u r e s e l e c t i o n procedure w i t h the d i s c r i m i n a n t f u n c t i o n a n a l y s i s . The purpose of t h i s i s to reduce the number of features that need to be used f o r object c l a s s i f i c a t i o n . The c r i t e r i a f o r feature s e l e c t i o n are t y p i c a l l y based on a n a l y s i s - o f - v a r i a n c e s t a t i s t i c s . Feature s e l e c t i o n i s performed i n a stepwise f a s h i o n , w i t h features being entered or removed one at a time on the b a s i s of the values c a l c u l a t e d f o r the s e l e c t i o n parameters. Various parameters and stepping procedures have been developed f o r the purposes of feature s e l e c t i o n , most of them w i t h the l i n e a r d i s c r i m i n a n t f u n c t i o n i n mind. The BMDP s t a t i s t i c a l package was used f o r the stepwise d i s c r i m i n a n t f u n c t i o n a n a l y s i s performed i n t h i s t h e s i s . B r i e f l y , the f e a t u r e s e l e c t i o n procedure i n t h i s program i s c a r r i e d out by e n t e r i n g features i n t o the a n a l y s i s according to which has the highest c o n d i t i o n a l F r a t i o . The c o n d i t i o n a l F r a t i o i s c a l c u l a t e d f o r each unentered fea t u r e through a one-way a n a l y s i s of covariance, where the c o v a r i a t e s are the features already entered ( f o r the f i r s t step, where no features have yet been entered, the feature w i t h the highest value f o r the F s t a t i s t i c i s entered). The F s t a t i s t i c i s a measure of the d i f f e r e n c e s between group means. S i m i l a r l y , the c o n d i t i o n a l F r a t i o used i n subsequent steps i s a measure of the c o n t r i b u t i o n of the feature i n question to group d i f f e r e n c e s , given the features already included. 238 Features are entered i n t o the a n a l y s i s one at a time, u n t i l no more of them s a t i s f y a u s e r - s p e c i f i e d minimum F-value. Features that are h i g h l y c o r r e l a t e d w i t h those already entered i n t o the a n a l y s i s w i l l not, however, be entered. Features may a l s o be removed at each step, i f , by t h e i r removal, the Wilks' A s t a t i s t i c i s lower than i t was i n previous steps which had the same number of features already entered. W i l k s ' A i s a m u l t i v a r i a t e a n a l y s i s of variance s t a t i s t i c t hat t e s t s the e q u a l i t y of group means using the s e l e c t e d f e a t u r e s . Once the f i n a l set of features has been s e l e c t e d , the l i n e a r d i s c r i m i n a n t f u n c t i o n i s c a l c u l a t e d according to Equation 10.7. 10.2 Linear-Quadratic Models of R a d i a t i o n A c t i o n Of the mathematical equations used to describe mammalian c e l l s u r v i v a l curves, among the most commonly used i s the s o - c a l l e d l i n e a r - q u a d r a t i c equation: S = e-e<D> = e-aD-P°2 (10.8) where D i s the dose, S i s the s u r v i v i n g f r a c t i o n , and e(D) i s the number of l e t h a l l e s i o n s produced as a f u n c t i o n of dose. While there are b i o p h y s i c a l reasons f o r the use of such an equation, i t a l s o represents the f i r s t few terms of a Taylor s e r i e s expansion about zero dose f o r any a r b i t r a r y f i t t i n g f u n c t i o n ( i . e . e(D) = a 0 + a tx + a 2 x 2 + . . . , w i t h a 0 = 0 s i n c e no r a d i a t i o n i n j u r y occurs at zero dose). Equation 10.8 would the r e f o r e be expected to f i t s u r v i v a l data reasonably w e l l at low doses, even i f i t d i d not represent the a c t u a l s u r v i v a l dose - response. 239 The exponential dependence of c e l l s u r v i v a l on e(D) i n Equation 10.8 i s a consequence of the f a c t that the p r o b a b i l i t y of a c e l l i n an i r r a d i a t e d p o p u l a t i o n r e c e i v i n g a l e t h a l i n j u r y i s determined by Poisson s t a t i s t i c s . The s u r v i v i n g f r a c t i o n at a s e l e c t e d r a d i a t i o n dose i s given by the p r o b a b i l i t y that a c e l l has r e c e i v e d no l e t h a l l e s i o n s ( i . e . p(e-0) = e " € ) . The concept of s u b l e s i o n i n t e r a c t i o n to produce a l e t h a l i n j u r y to the c e l l i s g e n e r a l l y invoked to produce the l i n e a r and qu a d r a t i c terms seen i n Equation 10.8. According to t h i s concept, " s u b l e s i o n s " produced by the i o n i z a t i o n s along a r a d i a t i o n t r a c k are not i n themselves l e t h a l , but p a i r w i s e i n t e r a c t i o n of sublesions can r e s u l t i n a " l e t h a l " l e s i o n . The i n t e r a c t i n g sublesions may be the products of the i o n i z a t i o n s along a s i n g l e p a r t i c l e t r a c k ( y i e l d i n g the l i n e a r dose-dependence), or may r e s u l t from i o n i z a t i o n s along two independent t r a c k s (producing the quadratic dose dependence). The p o s s i b l e existence of such mechanisms was f i r s t proposed on the b a s i s of chromosome aberrations ( p a r t i c u l a r l y chromosome exchanges) observed i n c e l l s a f t e r r a d i a t i o n exposure (Lea, 1955). This type of damage re q u i r e s breaks to occur i n two separate chromosomes, and so can r e a d i l y be r e l a t e d to the concept of s u b l e s i o n i n t e r a c t i o n . However, extension of these ideas to more complex e f f e c t s such as c e l l k i l l i n g r e q u i r e s more general assumptions to be made about the r e l a t i o n s h i p between energy d e p o s i t i o n events and the observed e f f e c t . Paramount among these i s , of course, the assumption that the magnitude of the f i n a l e f f e c t i s d i r e c t l y p r o p o r t i o n a l to the degree of the i n i t i a l damage. The b a s i s f o r suggesting a l i n e a r - q u a d r a t i c dose-response f o r c e l l k i l l i n g has i n one case (Chadwick and Leenhouts, 1973) been the s u p p o s i t i o n that the l e t h a l l e s i o n i s a complete break i n some segment of the c e l l ' s nuclear DNA. This r e q u i r e s that both strands of the double-stranded DNA 240 molecule be broken, thus suggesting the existence of both one-track and two-track modes of l e s i o n production. Another l i n e a r - q u a d r a t i c model developed to describe r a d i a t i o n dose-response i n b i o l o g i c a l systems i s the Theory of Dual R a d i a t i o n A c t i o n (TDRA). This model, developed by K e l l e r e r and R o s s i (1972), attempts to r e l a t e m i c r o d o s i m e t r i c a l q u a n t i t i e s to a v a r i e t y of observable r a d i a t i o n e f f e c t s such as chromosome a b e r r a t i o n s , mutation, t i s s u e damage, and c e l l k i l l i n g . The nature of the sublesions i s not s p e c i f i e d i n t h i s model, but they are assumed to be formed w i t h i n s m a l l , s p h e r i c a l " s e n s i t i v e s i t e s " w i t h a y i e l d p r o p o r t i o n a l to the energy imparted per s i t e . The r e l e v a n t measure of the energy imparted i s p o s t u l a t e d to be the s p e c i f i c energy, z, imparted to a reference volume that corresponds to the s i z e of the s e n s i t i v e s i t e . Values of z f o r a given r a d i a t i o n modality and dose are subject to s t a t i s t i c a l v a r i a t i o n s from s i t e to s i t e , and must ther e f o r e be described by a p r o b a b i l i t y d i s t r i b u t i o n f(z;D) (where D i s the t o t a l dose imparted to the system). The f u n c t i o n f(z;D) depends on the shape and s i z e of the reference volume, as w e l l as on r a d i a t i o n q u a l i t y , but i t s mean value i s equal to D: z = z f(z;D) dz = D. (10.9) Through the concept of s u b l e s i o n i n t e r a c t i o n , the y i e l d of l e t h a l l e s i o n s , e(D), i s p r o p o r t i o n a l (with p r o p o r t i o n a l i t y constant, k) to the square of the s p e c i f i c energy: 241 k z 2 f(z;D) dz = k z 2(D) (10.10) I t was shown by K e l l e r e r and R o s s i th a t : z 2(D) = r D + Z)2 (10.11) thus l e a d i n g to the one-track and two-track components of l e s i o n production. The c o e f f i c i e n t J" i s defined i n terms of the d i s t r i b u t i o n of s p e c i f i c energies, f1(z), a r i s i n g from s i n g l e events (a s i n g l e event i s defined as energy d e p o s i t i o n i n a s i t e by a s i n g l e i o n i z i n g p a r t i c l e and/or i t s secondaries): f = Z i 2 A i z 2 f x ( z ) dz z f i ( z ) dz. (10.12) f i s r e f e r r e d to as the energy average of the event s i z e , and can be determined from microdosimetric measurements (e.g. E l l e t t and Braby, 1972). For c e l l s u r v i v a l , the TDRA becomes: = *-e0» = p-k(CD+D*) (10.13) and i t s c o e f f i c i e n t s can be e a s i l y defined i n terms of the constants a and /3 more commonly seen i n the r a d i o b i o l o g i c a l l i t e r a t u r e : 242 a = k r 0 = k r = OC/B. (10.14) (10.15) (10.16) Since i t s i n i t i a l development, the TDRA has been modif i e d i n order to el i m i n a t e some of the s i m p l i f y i n g assumptions that had been a p p l i e d to the o r i g i n a l f o r m u l a t i o n . The s o - c a l l e d Generalized Theory of Dual R a d i a t i o n A c t i o n (GTDRA) ( K e l l e r e r and R o s s i , 1978) deals more r i g o r o u s l y w i t h the treatment deals w i t h the p r o b a b i l i t y of s u b l e s i o n i n t e r a c t i o n f o r a given s e p a r a t i o n d i s t a n c e , which had been assumed to be constant over the dimensions of the " s e n s i t i v e s i t e " (and zero outside i t ) i n the o r i g i n a l TDRA. Nevertheless, the r e s u l t i n g form of the equation d e s c r i b i n g the dose-response r e l a t i o n i s s t i l l l i n e a r - q u a d r a t i c : The c o e f f i c i e n t £ cannot, however, be d i r e c t l y determined from microdosimetric measurements, as was the case f o r C. This i s because £, f i r s t of a l l , i s considered to be dependent on energy d e p o s i t i o n p a t t e r n s , t( x ) , i n volumes of nanometer ( r a t h e r than micrometer) dimensions, which are too small to be measured d i r e c t l y . Second, £ i s a l s o dependent on two b i o l o g i c a l parameters: the geometry of the s e n s i t i v e t a r g e t (described by the f u n c t i o n s ( x ) , which i s a measure of the t a r g e t volume at a distance x from a s u b l e s i o n ) , and the p r o b a b i l i t y , g(x), that two sublesions separated by distance x i n t e r a c t : geometry of the s e n s i t i v e regions i n the c e l l . Most notably, t h i s (10.17) 243 s ( x ) g ( x ) t ( x ) hnpx2 dx s(x)g(x) dx = t ( x ) 7 ( x ) dx (10.18) The p r o p o r t i o n a l i t y constant, k, i s a l s o dependent on s(x) and g(x), but i s s t i l l assumed to be independent of r a d i a t i o n q u a l i t y under most circumstances. The f u n c t i o n d e s c r i b i n g the b i o l o g i c a l q u a n t i t i e s c o n t r i b u t i n g to | cannot be d i r e c t l y measured, but i t has been suggested that i t can be der i v e d f o r a given b i o l o g i c a l system from measurements of £ = a/fl made f o r a range of LET values (Zaider and R o s s i , 1985), or perhaps even from m i c r o d o s i m e t r i c a l measurements made i n a s e r i e s of s p h e r i c a l s i t e s of d i f f e r e n t dimensions (Zaider and R o s s i , 1988). S i m i l a r l y , methods of c a l c u l a t i n g energy d i s t r i b u t i o n s f o r nanometer-sized volumes on the b a s i s of microdosimetric measurements i n micrometer-sized volumes have been proposed (Zaider et a l , 1982), as have methods f o r determining these q u a n t i t i e s through Monte Carlo s i m u l a t i o n s of charged p a r t i c l e t r a c ks (Brenner and Zaider, 1984). 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0084995/manifest

Comment

Related Items