Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A computer visual-input system for the automatic recognition of blood cells Cossalter, John George 1970

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1970_A7 C68.pdf [ 6.61MB ]
Metadata
JSON: 831-1.0102199.json
JSON-LD: 831-1.0102199-ld.json
RDF/XML (Pretty): 831-1.0102199-rdf.xml
RDF/JSON: 831-1.0102199-rdf.json
Turtle: 831-1.0102199-turtle.txt
N-Triples: 831-1.0102199-rdf-ntriples.txt
Original Record: 831-1.0102199-source.json
Full Text
831-1.0102199-fulltext.txt
Citation
831-1.0102199.ris

Full Text

A COMPUTER VISUAL-INPUT SYSTEM FOR THE AUTOMATIC RECOGNITION OF BLOOD CELLS by JOHN GEORGE COSSALTER B.A.Sc, The U n i v e r s i t y o f B r i t i s h C o l u m b i a , 1968 A THESIS SUBMITTED I N PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE i n t h e Department o f E l e c t r i c a l E n g i n e e r i n g We a c c e p t t h i s t h e s i s as c o n f o r m i n g t o the r e q u i r e d s t a n d a r d R e s e a r c h S u p e r v i s o r , Members o f the Committee Head o f the Department Members o f the Department o f E l e c t r i c a l E n g i n e e r i n g THE UNIVERSITY OF BRITISH COLUMBIA A u g u s t , 1970 In p r e s e n t i n g t h i s t h e s i s in p a r t i a l f u l f i l m e n t o f t h e r e q u i r e m e n t s f o r an a d v a n c e d d e g r e e a t t h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , I a g r e e t h a t t h e L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r a g r e e t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y p u r p o s e s may be g r a n t e d by t h e Head o f my Depar tment o r by h i s r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Depar tment o f The U n i v e r s i t y o f B r i t i s h C o l u m b i a V a n c o u v e r 8, Canada ABSTRACT A computer v i s u a l - i n p u t system was b u i l t f o r the purpose of studying the c l a s s i f i c a t i o n of leukocytes. I t consisted of an image d i s s e c t o r camera i n t e r f a c e d d i r e c t l y to a D.E.C. PDP-9 computer; a d i s p l a y of the image f i e l d was also provided, using a monitoring scope. The design and hardware arrangement of the system i s b r i e f l y des-cribed, while d e t a i l e d diagrams of the l o g i c networks are shown i n Appendix I I . Photomicrographs of neutrophils were used as a pattern set, i n a study of the computer c l a s s i f i c a t i o n of c e l l age and l o b u l a r i t y . C l u s t e r i n g of feature vectors was noted i n a two-dimensional measurement space showing that metamyelocyte, banded and segmented c e l l s can be distinguished. A square contour-trace of the neutrophil n u c l e i was performed and an area operator pre-processed the shape of a nucleus i n t o a curvature function. Peaks i n t h i s curvature function, a measure of l o b u l a r i t y , as well as the r a t i o of the •perimeter to square root of nuclear area, a measure of the i r r e g u l a r i t y i n the nuclear boundary, were used as o r i e n t a t i o n and size-independent features. The area operator was found to be unsuitable f o r e x t r a c t i n g curvature from leukocyte images. In cases of extreme nuclear curvature and nuclear filamentation, the basic formulations of the operator were v i o l a t e d g i v i n g an erroneous measure of curvature. The general form of the frequency spectrum of the video s i g n a l from the image d i s s e c t o r camera was derived. The s i g n a l bandwidth requirements and the camera r e s o l u t i o n were found experimentally. i i TABLE OF CONTENTS Page L i s t of I l l u s t r a t i o n s i v Acknowledgement v i 1. Introduction 1 1.1 The D i f f e r e n t i a l Leukocyte Count 1 1.2 The Leukocyte Smear 3 1.3 The C l a s s i f i c a t i o n of Leukocytes 4 2. The Computer Visual-Input System 7 2.1 Methods of Picture Processing 7 2 .2. The Design of the System 8 2.3 The Display Monitor . . . . . 12 2.4 The Computer Interface • 13 3. The System Software 15 3.1 Test Programs 15 3.2 The Simulation of the Area Operator, and Ridge Function . . . . 16 3.3 Problems Encountered i n Computing the Ridge Function 19 4. A Neutrophil C l a s s i f i c a t i o n Study 23 5. The System Parameters . 29 5.1 Noise C h a r a c t e r i s t i c s of the Image Dissector Tube 29 5.2 The S p a t i a l Frequency Spectrum 31 5.3 The S p a t i a l Resolution 34 5.4 Bandwidth Requirements 36 6. Conclusions • • • 38 APPENDIX I 40 APPENDIX II 42 APPENDIX I I I 52 BIBLIOGRAPHY 53 i i i LIST OF ILLUSTRATIONS Figure Page 1.1 A d i f f e r e n t i a l leukocyte count examination form 2 1.2 Photomicrograph of a neutrophil surrounded by erythrocytes (2000 X ) : 4 1.3 The area operator positioned at the point s 1 on the curve C 5 1.4 A tes t pattern and i t s ridge f u n c t i o n • 6 2.1 Block diagram of the v i s u a l - i n p u t system 10 2.2 A p i c t o r i a l diagram of the transducer 11 2 .3 I n s t r u c t i o n s executable on the v i s u a l - i n p u t device . . . . 14 3 . 1 The g r e y - l e v e l c r o s s - s e c t i o n of a neut r o p h i l 16 3.2 The contour-path and ridge f u n c t i o n of a neutrophil nucleus 19 3 .3 Contour-traces showing the e f f e c t s of i n c r e a s i n g the g r i d spacing . . . . . . . . . 20 3 . 4 T y p i c a l errors made when operating on a c t u a l n e u t r o p h i l n u c l e i 22 3 .5 Ridge functions derived using d i f f e r e n t area operator s i z e s 23 4 .1 The general nuclear shapes of neutrophils 24 4 . 2 T y p i c a l ridge functions 25 4 . 3 A measurement-space graph showing the c l u s t e r i n g of neu t r o p h i l age groups 27 5.1 A model of the degradation of an image 34 5.2 A scan across the edge of a razor blade 36 i v Figure Page A2.1 Device co n t r o l 43 A2.2 Mode co n t r o l 44 A2.3 Data t r a n s f e r c o n t r o l 45 A2.4 A/D converter 46 A2.5 8 b i t X (Y) r i p p l e counter 47 A2.6 X (Y) load pulse c o n t r o l . . 48 A2.7 X (Y) attenuator 49 A2.8 X (Y) D/A converter . . . 50 A2.9 X (Y) data gates 51 A3.1 Schematic of the logarithmic a m p l i f i e r 52 v ACKNOWLEDGEMENT I would l i k e to g r a t e f u l l y acknowledge the National Research Council f o r t h e i r f i n a n c i a l support through a studentship i n 1968-69 and the research grant 67-3350 i n 1969-70. Acknowledgement i s also given to my supervisor, Dr. J. S. MacDonald, f o r h i s support and guidance, and to Dr. R. ¥. Donaldson and Dr. R. H. Pearce f o r t h e i r reading and constructive c r i t i c i s m of the thesi s . I would also l i k e to thank Mr. A. Leugner f o r h i s e f f i c i e n t and expert construction of the hardware, and Mr. B. J. Twaites and Mr. H. H. Black f o r invaluable assistance i n obtaining photographs and s l i d e s . Thanks are also due to Heather DuBois f o r typing the manuscript and my fr i e n d s and colleagues f o r proofreading. v i 1 1. INTRODUCTION 1.1 The D i f f e r e n t i a l Leukocyte Count The d i f f e r e n t i a l leukocyte (white blood c e l l ) count i s a routine and important diagnostic test used i n c l i n i c a l medicine. The process, which con s i s t s of tab u l a t i n g the r e l a t i v e frequency of the d i f f e r e n t types of leukocytes, i s invaluable i n i n d i c a t i n g a large v a r i e t y of human disorders, such as acute i n f e c t i o n s , i n t o x i c a t i o n s or drug reactions, malignancies and severe hemorrhages. A sample of peripheral blood i s taken, smeared on a glass s l i d e and allowed to dry. The s l i d e i s then commonly treated with Wright's s t a i n , to chemically dye the c e l l s according to t h e i r i o n i c a f f i n i t y , and examined under a microscope. Usually, one hundred c e l l s are s c r u t i n i z e d and classed i n t o standard categories, each category having s i m i l a r morphological properties which define i t s topology and b i o l o g i c a l structure. Nuclear and cytoplasmic shape, the presence of a n u c l e o l i , size,- texture and color are t y p i c a l properties used to c a l c u l a t e a percentage l i s t of c e l l types. (Figure 1.1 shows a t y p i c a l form, used i n a c l i n i c a l laboratory, to report the "normality" of a blood smear). The f o l l o w i n g f a c t s suggest that the d i f f e r e n t i a l leukocyte count, to some degree, should be automated: a) Aside from being a tedious task, the process i s an expensive c l i n i c a l t est, considering the cost of the technician's time to prepare and examine a blood smear. b) A s i g n i f i c a n t portion, about 10$, of the smears processed i n a c l i n i c a l laboratory are abnormal and are brought to the a t t e n t i o n of a hematologist or experienced s p e c i a l i s t f o r a more d e t a i l e d examination. 2 UNIVERSITY OF TENNESSEE H E M A T O L O G Y LABORATORY MEMPHIS, TENNESSEE EXAMINATION OE PERIPHERAL IiLOOD Nnme . .Date . . . Age M F W C Referred by Red Blood Cells: Thrombocytes: White Blood Cells: Number counted Norma Per Cent Stem Cell Myeloblast Progranulocyte N . Myelocyte 0-1 N. Metamyelocyte 2-10 N. Hand 50-70 N. Segmented 1-4 Eosinophil 0-1 Basophil Lymphoblast Prolymphocyte 20-40 Lymphocyte 1-6 Monocyte Plasmocyte Atypical cell Disintegrated cells per 100 intact VVBC Summary of Abnormalities: Interpretation: Examined by, Figure 1.1 A d i f f e r e n t i a l leukocyte count examination form. 3 c ) There w i l l be a s t r o n g need i n the f u t u r e f o r a f u l l y automated c l i n i c a l s c r e e n i n g l a b o r a t o r y w i t h i n w h i c h an i m p o r t a n t sub-s y s t e m i s t h e d i f f e r e n t i a l l e u k o c y t e c o u n t . d) W i t h t h e a d v e n t o f l a r g e - s c a l e d i g i t a l computers and t h e i r i n c r e a s e d u s e i n p a t t e r n c l a s s i f i c a t i o n , the means a r e a t hand t o c l a s s l e u k o c y t e s q u a n t i t a t i v e l y , on t h e b a s i s o f t h e i r morpho-l o g i c a l c h a r a c t e r i s t i c s . The d i f f e r e n t i a l c o u n t i s n o t a d a p t a b l e t o s i m p l e methods o f a u t o m a t i o n as i s t h e e n u m e r a t i o n o f l e u k o c y t e s , e r y t h r o c y t e s ( r e d b l o o d c e l l s ) and t h r o m b o c y t e s ( s m a l l c y t o p l a s m f r a g m e n t s w h i c h a r e i n v o l v e d i n t h e sequence o f b l o o d c l o t t i n g ) . .The p r o c e s s r e q u i r e s complex and e x p e n s i v e p e r i p h e r a l d e v i c e s , s u c h as a r e m o t e l y c o n t r o l l e d m i c r o s c o p e [ l ] w i t h a means f o r image i n p u t t o a computer [ l ] - [ 7 ] , a computer and complex s o f t w a r e a l g o r i t h m s f o r c e l l c l a s s i -f i c a t i o n [ 2 ] , [4], [ 5 ] , [8], [ 9 ] . 1.2 The L e u k o c y t e Smear Human p e r i p h e r a l b l o o d i s composed o f t h r e e b a s i c c e l l s - l e u k o c y t e s , e r y t h r o c y t e s and t h r o m b o c y t e s suspended i n a medium o f plasma. The l e u k o c y t e s can be d i v i d e d i n t o f i v e d i s t i n c t m o r p h o l o g i c a l t y p e s : a) Lymphocytes b) Monocytes c) B a s o p h i l s d) E o s i n o p h i l s e) N e u t r o p h i l s There a r e f r o m 5000 t o 9000 l e u k o c y t e s per c u b i c m i l l i m e t e r , r a n g i n g f r o m 5 t o 50^u ( m i c r o n s ) i n s i z e . N e u t r o p h i l s , w h i c h a r e t h e f o c a l p o i n t o f t h i s t h e s i s , r ange f r o m 10 t o 15jU. The e r y t h r o c y t e s outnumber l e u k o c y t e s by a f a c t o r o f 600:1 and have an a v e r a g e s i z e o f a b out Su. 4 A t y p i c a l leukocyte as i t appears on a smear, shown i n Figure 1.2, i s not s p h e r i c a l as found i n c i r c u l a t i n g blood, but f l a t t e n e d i n t o a d i s c . The nucleus of the c e l l may be folded upon i t s e l f and the cytoplasm i s probably touched, or overlapped by, erythrocytes. Figure 1.2 Photomicrograph of a neutrophil surrounded by erythrocytes (2000x). 1.3 The C l a s s i f i c a t i o n of Leukocytes Looking at Figure 1.2, the problem of f i n d i n g good algorithms to c l a s s i f y leukocytes becomes apparent. F i r s t , the leukocyte must be found i n the image f i e l d . This i s not a great problem since l o c a t i n g the o p t i c a l l y dense nucleus locates the c e l l . Bourk [lO] has solved t h i s problem adequately. Once the leukocyte i s located, quantitative measurements must be made to c l a s s i f y i t morphologically. Folding may obscure the true t o p o l o g i c a l nature of the nucleus and touching erythrocytes may complicate the detection of the cytoplasmic boundary. (The software problem of the detection of overlapping 5 c e l l s has been solved by R i n t o l a and Hsu [ l l ] ) . • C l e a r l y , the c l a s s i f i c a t i o n of leukocytes must be made using features which are orientation-independent, such as, geometric nuclear shape. This thesis i s p r i m a r i l y concerned with the form or shape of the n u c l e i of neutrophils. The n u c l e i are contour-traced. An "area operator", based on the theory formulated by Connor [ l 2 ] , converts the shape of the contour into a "ridge function". Consider the curve C, i n Figure 1.3, to be the boundary between a l i g h t and a dark region. The area operator i s a d i s c of .area A^, having i t s center always on the curve C . A ridge f u n c t i o n V^(s), a f u n c t i o n of arc length, i s formed as the operator moves i n one d i r e c t i o n around the curve C. The curve C must be continuous with-i i n the operator d i s c and at each point s , the area operator i s defined to have a value V I 1 = A - A ' ( l . l ) r |s d H |s v I I 1 = the value of the ridge f u n c t i o n at s . A 1 = the area of the operator which l i e s i n i the l i g h t region, evaluated at s . From the d e f i n i t i o n of V ( s ) , r A d = — when C i s a s t r a i g h t l i n e . (l.2) A d O^V ^ — when C i s concave with respect to the r 2 dark region. ( l « 3 ) A d 7~<V <A, when C i s convex with respect to the 2 r d dark region. (l » 4 ) I n t u i t i v e l y , i t i s seen that V (s) w i l l give some measure of the shape of the curve. In f a c t , V (s) i s a non-linear f u n c t i o n of the curvature, K(s), of C; r V (s) becomes proportional to K(s) i n the l i m i t of small K. An important f a c t , r proven i n Connor's t h e s i s , which has become u s e f u l i n feature extraction, i s that the extrema of V (s) coincide with the extrema of K(s), both functions having the same o r i g i n and d i r e c t i o n along C. Thus, points of maximum and minimum curvature can be determined using the area operator. Figure 1.4 shows the ridge f u n c t i o n of a two-lobed f i g u r e . two-lobed pattern ridge function Figure 1.4 A tes t pattern and i t s ridge function. 7 This t h e s i s presents an explanation of peripheral hardware and the operation of the computer v i s u a l - i n p u t system which w i l l be used f o r the study of pattern c l a s s i f i c a t i o n of leukocytes. An i n v e s t i g a t i o n of the f e a s i b i l i t y of using the area operator as a preprocessing technique f o r future work, was used as a t e s t of the performance of the system. A r e a l pattern set of leukocyte transparencies was used i n a c l a s s i f i c a t i o n study. The study i n d i c a t e s that metamyelocyte, banded and segmented neutrophils can be well defined i n measurement space, and thus c l a s s i f i e d . The shape of n u c l e i was found to be an important d e s c r i p t i v e feature and should be f u r t h e r explored.- An evaluation of the system's p h y s i c a l parameters i s made and the advantages, of using a l a r g e r aperture image d i s s e c t o r tube are o u t l i n e d i n Appendix I. 2. THE COMPUTER VISUAL-INPUT SYSTEM 2.1 Methods of Picture Processing P i c t u r e processing by a computer e n t a i l s d i g i t i z i n g a p i cture frame i n t o an array of m by n g r i d points. Each of the mxn points i s interrogated by some sort of scanning instrument; the o p t i c a l density value of a point i s quantized, d i g i t a l l y coded and read into the computer. The processing of t h i s p i c t ure data requires that the computer have random access to any g r i d point, since i t i s assumed that there i s no a p r i o r i knowledge of the information contained i n the p i c t u r e . The f a s t e s t and most c o s t l y method of processing a p i c t u r e i s to s y s t e m a t i c a l l y read the o p t i c a l d e n s i t i e s of the mxn points d i r e c t l y into the. high-speed memory of a computer, where each point can be randomly accessed. Ledley [2] used a f l y i n g spot scanner to read a r a s t e r of 700 by 500 points t h r o u g h a d a t a c h a n n e l i n t o an IBM 360/44 sy s t e m ' s c o r e memory. H i s work on monocyte c l a s s i f i c a t i o n was done t o t a l l y w i t h i n t h e computer a t speeds l i m i t e d o n l y by t h e c o r e memory's c y c l e t i m e . I f t h e i n f o r m a t i o n c o n t a i n e d i n t h e mxn p o i n t s i s t o o l a r g e f o r t h e computer's h i g h - s p e e d memory, t h e d a t a c a n be s y s t e m a t i c a l l y r e a d o n t o an e x t e r n a l memory s u c h as m a g n e t i c t a p e . A l t h o u g h d a t a a c q u i s i t i o n f r o m s u c h a memory i s s l o w e r t h a n t h a t d i s c u s s e d above, t h i s method has t h e advantage t h a t t h e p i c t u r e may b e . p r o c e s s e d a t a l a t e r d a t e , i n d e p e n d e n t o f t h e s c a n n i n g i n s t r u m e n t . Mendelsohn e t a l [ 8 ] p r o c e s s e d a c t u a l l e u k o c y t e images. A f l y i n g s p o t s c a n n e r was o p e r a t e d d i r e c t l y t h r o u g h a m i c r o s c o p e , w r i t i n g p i c t u r e i n f o r m a t i o n o n t o m a g n e t i c t a p e . The d i s c r i m i n a t i o n o f t h e f i v e t y p e s o f l e u k o c y t e s was t h e n done f r o m t h e m a g n e t i c t a p e , u s i n g a H o n e y w e l l 800 computer. The s l o w e s t b u t p o t e n t i a l l y l e a s t c o s t l y method o f p i c t u r e p r o c e s s i n g i s t o u se a s c a n n i n g i n s t r u m e n t o n - l i n e w i t h a computer. T h i s t y p e o f s y s t e m employs t h e a c t u a l p i c t u r e as a memory, u s i n g t h e s c a n n i n g i n s t r u m e n t as a random-access t r a n s d u c e r . R o s e n b e r g ' s s y s t e m [ 5 ] employed a f l y i n g s p o t s c a n n e r i n t e r f a c e d t o a PDP-9 computer. P i c t u r e g r i d p o i n t s s p e c i f i e d by t h e PDP-9 were i n t e r r o g a t e d one a t a t i m e , o n - l i n e . 2.2 The D e s i g n o f t h e System There a r e two f a c t o r s w h i c h make d i r e c t r e a d - i n o f p i c t u r e d a t a t o a memory d e s i r a b l e : a) F u r t h e r p r o c e s s i n g o f t h e p i c t u r e may be done a t a h i g h speed, n o t h a v i n g t o communicate w i t h t h e s l o w p e r i p h e r a l p i c t u r e s c a n n i n g a p p a r a t u s , and b) Subsequent i n t e r r o g a t i o n s o f t h e same g r i d p o i n t t h r o u g h a t r a n s -d u c e r may g i v e d i f f e r e n t o p t i c a l d e n s i t y v a l u e s . P o s i t i o n j i t t e r . 9 and i l l u m i n a t i o n f l u x v a r i a t i o n s are inherent to picture scanning instruments and thus the same g r i d point may have a va r i a b l e o p t i c a l density value. Once the o p t i c a l density of a point has been written i n t o a memory, though, f u r t h e r i n t e r r o g a t i o n s of that point w i l l be the same. A d i r e c t picture-data read-in i s desirable, but i n most cases a large high-speed memory u n i t i s too expensive f o r the average h o s p i t a l c l i n i c a l laboratory. The system to be described i n t h i s t h e s i s has a f a s t e r random access time than e i t h e r a d i s c or drum memory, and on-line processing allows the operator to observe, on a monitoring scope, i f the system i s f u n c t i o n i n g properly. . The unique properties of the image d i s s e c t o r camera make i t adaptable f o r i n t e r f a c i n g with a small computer such as the PDP-9, using the actual picture as a random-access "read-only" memory. This type of video camera requires no image storage and thus can be operated at a v a r i a b l e scan rate without a change i n the s i g n a l current amplitude; i n f a c t , the scan can be stopped completely. The d i r e c t i o n of scan i s also v a r i a b l e allowing the FDP-9 to interrogate any g r i d point randomly. Figure 2.1 shows a block diagram of the computer v i s u a l - i n p u t system to be used f o r leukocyte c l a s s i f i c a t i o n studies. The system i s set up to provide a f l e x i b l e means of processing the information on transparencies, using the camera as the scanning device and simultaneously viewing the f i e l d of scan on a monitoring scope. The di s p l a y makes i t possible to monitor and detect processing e r r o r s and i n i t i a l l y a l i g n the scan f i e l d of i n t e r e s t . The system hardware which i s b u i l t up mainly from D.E.C. modules, can be divided i n t o two main sections: a) the i n t e r f a c i n g hardware which provides c o n t r o l commands and data POP-9 PWR CLR. IOP. DEVICE & SUBDEVICE SELECTION I? A V I/O SKIPRQ. DEVICE CONTROL A V 2-CLOCK STOP. CLOCK STRT. LOAD X LOAD Y CLEAR SYSTEM -TFSTART A/D CONVERSION CLEAR A/D FLAG ENABLE DATA LINES READ A/D 18 DATA LINES READ RO. AA> CONVERSION COMPLETE A/D CONVERTER ~J^D A/D DIGITAL OUTPUT 8 BIT X RIPPLE COUNTER CLOCK OUTPUT (X TOGGLE) SX COUNTER f OUTPUT X DATA GATES MODE CONTROL D/A INPUT CONTROL 1 DATA TRANSFER CONTROL 10 DATA LINES Y DATA GATES Y COUNTER TOGGLE 1 DATA LINES 5 3 x D/A CONVERTER LOAD X DATA LOAD Y PULSE X LOAD PULSE CONTROL X D/A OUTPUT. ATTENUATOR DISPLAY X DEFLECTION DISPLAY UNIT DISPLAY INPUT LOW-PASS FILTER LOW-PASS FILTER J LOG. ™AMP. CAMERA X DEFLECTION IMAGE DISSECTOR CAMERA VIDEO SIGNAL DISPLAY Y DEFLECTION S DATA LINES 0 Y D/A CONVERTER Y D/A OUTPUT , ATTENUATOR CAMERA Y DEFLECTION LOAD Y DATA Y LOAD PULSE CONTROL Y COUNTER OUTPUT 8 BIT Y RIPPLE COUNTER FIGURE 2.1 BLOCK DIAGRAM OF THE VISUAL-INPUT SYSTEM. O 11 t r a n s f e r between the camera and the PDP-9, and b) the monitoring equipment which provides a t e l e v i s i o n - t y p e image of the f i e l d of scan. Photomicrographs of leukocytes to be examined are made int o 2" x 2" transparent s l i d e s . Each s l i d e i s placed i n a s l i d e copier and i l l u m i n a t e d by a d.c. lamp; an o p t i c a l image of the s l i d e i s focused onto the photosensitive cathode of the camera. (See Figure 2.2) lamp s l i d e image d i s s e c t o r tube lens A photocathode magnetic focusing • aperture plane photomultiplier video a m p l i f i e r Figure 2.2 A p i c t o r i a l diagram of the transducer. The photocathode emits electrons with a density proportional to the in c i d e n t luminous f l u x . An a c c e l e r a t i n g p o t e n t i a l a t t r a c t s these electrons to the anode. Focusing i s accomplished, using a large a x i a l magnetic f i e l d . The r e s u l t i s an e l e c t r o n i c image of the o p t i c a l image, focused onto the aperture plane. Since the magnification of the image i n the d r i f t s e c t i o n of the image di s s e c t o r tube i s unity, an aperture plane v a r i a t i o n i n cathode current, I , i s given by the equation where S = the photocathode s e n s i t i v i t y . F = the i n c i d e n t luminous f l u x . A small aperture at the anode plane (a 1 mil (lO inches) diameter c i r c u l a r aperture i n t h i s case) allows a portion of the image current to impinge on the f i r s t dynode of a photomultiplier. This current i s then amplified by a f a c t o r of about 10^ and fed i n t o a video a m p l i f i e r to produce the video s i g n a l voltage. I t i s obvious, then, that the e l e c t r o n i c image i s l i t e r a l l y d issected by the aperture. The whole e l e c t r o n i c image i s d e f l e c t e d by a v e r t i c a l and a h o r i z o n t a l magnetic f i e l d so that any part of the image can be made to o v e r l i e the aperture i n t h i s way, any part of the image can be interrogated and the r e s u l t i n g s i g n a l current i s proportional to the transmittance of the s l i d e . Since the motion of the image and aperture are r e l a t i v e to one another, i t i s convenient to consider a moving aperture, scanning a s t a t i o n a r y image. 2.3 The Display Monitor To provide a d i s p l a y of the f i e l d to be processed, a 256 by 256 spot, v a r i a b l e frame rate, t e l e v i s i o n - t y p e scan i s used. The video s i g n a l from the image d i s s e c t o r camera i s low-pass f i l t e r e d to increase the signal-to-noise r a t i o and fed to the i n t e n s i t y input of a Tektronix 602 d i s p l a y u n i t . For a f i l t e r bandwidth of 20 kHz ( k i l o h e r t z ) the frame rate was a frame per second. (The long-persistence phosphor of the scope provided f o r a good display).. The d e f l e c t i o n of the camera and the scope are simultaneously c o n t r o l l e d by a 16 b i t r i p p l e counter. The f i r s t 8 b i t s of the counter are used as the h o r i z o n t a l d e f l e c t i o n c o n t r o l ; the l a t t e r 8 b i t s , the v e r t i c a l c o n t r o l . A v a r i a b l e speed clock toggles the r i p p l e counter, incrementing the e x i s t i n g binary number; the overflow from the h o r i z o n t a l part of the counter increments the v e r t i c a l part. The respective binary numbers are gated to the 8 most s i g n i f i c a n t b i t s of two 10 b i t D / A converters, which produce the d e f l e c t i o n voltages. The frame rate of the di s p l a y can be c o n t r o l l e d by adjusting the clock r e p e t i t i o n r a t e . (See Figure 2.1 and the Figures i n Appendix II f o r d e t a i l s ) . Only 256 increments of d e f l e c t i o n along the h o r i z o n t a l and v e r t i c a l axes are required to stay within the p r a c t i c a l r e s o l u t i o n l i m i t of the d i s p l a y u n i t . Using only 2"^ g r i d points f o r d i s p l a y also allows a f a s t e r frame rate, which i s s u b j e c t i v e l y a desi r a b l e feature f o r monitoring. The a c t u a l scan area on the photocathode i s a 0.8 inch square. There are 1024 by 1024 fundamental g r i d points within that area and the di s p l a y r a s t e r merely uses every fo u r t h point i n every f o u r t h g r i d point l i n e . 2.4 The Computer Interface With the i n t e r f a c e provided, the PDP-9 controls the mode of operation: a) In the di s p l a y mode, the external hardware i s permitted to d i s p l a y the f i e l d of scan. b) In the computer-read mode, the PDP-9 d e f l e c t s the camera to the g r i d coordinates i t s p e c i f i e s and reads the o p t i c a l density of that point on the transparency image. Although the PDP-9 i / o bus con s i s t s of 36 l i n e s , only those l i n e s required f o r c o n t r o l or data t r a n s f e r were i n t e r f a c e d to the external v i s u a l -input device. The 18 b i d i r e c t i o n a l data l i n e s were used f o r data tr a n s f e r ; a l l 18 l i n e s were wired from the PDP-9 to the device,for f l e x i b i l i t y , although only 10 are presently i n a c t i v e use. (10-bit d e f l e c t i o n coordinate codes are sent to the device and 6-bit o p t i c a l density codes are returned to the PDP-9 v i a the i / o bus data l i n e s ) . The three IOP l i n e s served to transmit i n s t r u c t i o n s to the external device and were properly gated f o r execution by the s i x DEVICE SELECTION and two SUBDEVICE SELECTION l i n e s , i / o SKIP and READ REQUEST int e r r u p t e d the 14 computer f o r asynchronous reading of o p t i c a l density codes. The table i n Figure 2.3 shows the i n s t r u c t i o n codes which co n t r o l the operation of the scanning device. They form part of the e s s e n t i a l software • f o r the system. Code Word i n Octal Operation Performed 707001 stop the clock and disable the h o r i z o n t a l counter gates 707002 load the i / o data l i n e s i n t o the h o r i z o n t a l D/A converter 707004 not used 707021 stop the clock and disable the v e r t i c a l counter gates 707022 , load the i / o data l i n e s i n t o the v e r t i c a l D/A converter 707024 not used 707041 s t a r t conversion of A/D 707042 s t a r t the external clock 707044 c l e a r the system 707061 check i f A/D conversion i s complete (i/O skip) 707062 . read the A/D code onto the i/O data l i n e s 707064 c l e a r the A/D conversion com-plete f l a g and reset the A/D converter Figure 2.3 Instructions executable on the v i s u a l - i n p u t device. The f u l l 10 b i t c a p a b i l i t y of the D/A converters, also employed by the d i s p l a y equipment, i s used f o r decoding the d e f l e c t i o n coordinates sent 20 by the PDP-9. The computer i s then able to interrogate any of the 2 points i n the scan area. A successive approximation A / D converter i s used to transform the continuous video s i g n a l voltage i n t o a 6 b i t binary number ranging from 0 to For c l e a r e r comprehension of the computer-read mode, the steps of a t y p i c a l data-fetch cycle are outlined. The p a r t i a l program shown below reads the o p t i c a l density of the g r i d point (x.,y.) into the P D P - 9 . R E A D 0 / E N T E R S U B R O U T I N E T O I N T E R R O G A T E A P O I N T 707001 / S T O P T H E C L O C K & I N T E R R U P T T H E D I S P L A Y L A C X I 707002 / T R A N S F E R X C O O R D I N A T E T O X D / A L A C Y I 707022 / T R A N S F E R Y C O O R D I N A T E T O Y D / A J M S D E L A Y / W A I T F O R F I L T E R S I G N A L D E L A Y - G O T O A P R O G R A M M E D D E L A Y 707041 / S T A R T A / D C O N V E R S I O N 707061 / I S P D P - 9 R E A D Y T O A C C E P T I N P U T D A T A ? J M P .-1 / N O , A S K A G A I N 707076 / Y E S , R E A D O P T I C A L D E N S I T Y C O D E I N T O P D P - 9 J M P * R E A D / E X I T F R O M S U B R O U T I N E D E L A Y 0 / A 100 M I C R O S E C O N D L O O P D E L A Y LAC(777740 DAC D# ISZ D JMP .-1 JMP*DELAY • I t should be noted that the delay can e i t h e r be a '•'do-nothing" loop, as shown, or, more e f f i c i e n t l y , the program can be rest r u c t u r e d so that the delay time i s taken up by some other i n s t r u c t i o n s which process data. 3. T H E S Y S T E M S O F T W A R E 3.1 Test Programs A program c a l l e d "TEST70" was written f o r the PDP-9 to check the operation of the computer-device i n t e r f a c e . Each of the commands tabled i n Figure 2.3 could be dispatched, upon teletype command, i n a loop so that the 16 execution and timing could be seen on an o s c i l l o s c o p e , by t r i g g e r i n g the scope on the proper IOP l i n e . This program f a c i l i t a t e s the checking of the proper timing sequences of a l l l o g i c a l functions performed by the system. But, to check the l i m i t a t i o n s of system parameters such as r e s o l u t i o n , a second major program c a l l e d "RID" was written; i t produced the ridge function discussed i n Chapter 1. The use of the c a l c u l a t e d ridge function i n the c l a s s i f i c a t i o n of neutrophils was also investigated. 3 . 2 The Simulation of the Area Operator and Ridge Function The f i r s t problem encountered i n implementing the ridge function as a descriptor of the shape of n u c l e i was to locate the leukocyte nucleus. For-tunately, the nucleus i s more opaque than any other part of the c e l l or surrounding erythrocytes and o p t i c a l density plots were made, using the picture input device, to t r y to define a s p e c i f i c g r e y - l e v e l which was c h a r a c t e r i s t i c of the nucleus. A t y p i c a l grey-level p r o f i l e of a neutrophil i s shown i n Figure 3 . 1 . plane of cross-section cytoplasm nucleus white black photomicrograph cross-section Figure 3 . 1 The grey-level cross-section of a neutrophil. The transmittance of the photomicrograph was quantized l i n e a r l y i n t o 8 l e v e l s and the nucleus of the neutrophil was found to have a g r e y - l e v e l of 0. (Only three b i t s of o p t i c a l density r e s o l u t i o n , over a dynamic range of 5 v o l t s , were used because of the signal-to-noise r a t i o l i m i t a t i o n s of the image d i s s e c t o r camera). To d i s t i n g u i s h a form of i n t e r e s t from the re s t of the picture, the program "RID" defines a g r e y - l e v e l window; o p t i c a l d e n s i t i e s l y i n g within t h i s window are c a l l e d "black" and those outside, "white". The window has a va r i a b l e width and v a r i a b l e o p t i c a l density p o s i t i o n . For instance, d e f i n i n g the l e v e l 0 to be black and others white, s u f f i c e s to d i s t i n g u i s h a black nucleus. I t i s also possible to d i s t i n g u i s h the cytoplasm of the neutrophil by having a window with g r e y - l e v e l s from 2 to 5 i n c l u s i v e . To locate the nucleus i n the scan f i e l d a t e l e v i s i o n - t y p e scan was used. S t a r t i n g at a programmed point near the lower l e f t corner of the image, a scan i s made from bottom to top, stepping i n increments of the g r i d point spacing, GS; i f a black point i s not found the scan i s reset to the bottom of the image and incremented GS h o r i z o n t a l l y . Upon l o c a t i n g the f i r s t black point, (x^,y^), the program switches from a scan to a contour-trace mode. At t h i s time, a neighbourhood around the point (x ,y ) i s defined by the regions x + 2GS and y ^ 2GS (3.1) A f t e r t h i r t y trace steps, i f a point i s found wi t h i n t h i s region, the trace mode w i l l be halted. One complete contour of a closed f i g u r e w i l l encounter one point within t h i s neighbourhood and h a l t the trace mode. (With j i t t e r noise i n the device i t cannot be ensured that the point (x-j_»y-j_) w i l l be en-countered again a f t e r contour-tracing the nucleus and thus a small neighbourhood of the s t a r t i n g point i s defined to stop the trace mode). For a black f i g u r e , the neutrophil nucleus i n t h i s case, and a white environment," a clockwise trace i s made around the f i g u r e i n a simple square step algorithm. The algorithm, as used by Mason and Clemens [l3], i s given by the f o l l o w i n g r u l e s : f i r s t x = x n (3.2) o 1 y 0 = y x - cs (3.3) and upon interrogating (x ,»y, ) and finding i t i s bla.ck: x k + i = x k + C y k - i " y k ] ( 3 ' 4 ) y-k+l or i f (x ,y ) i s a white point: + t\ ~ x k _ x ] (3-5) x _ = x k + [ y f e - y ^ ] (3.6) k+l = yk + [ V i " x k ] 'k+l and i f three white points or three black points are encountered i n succession, the next point i s treated as i f i t were black or white r e s p e c t i v e l y . I t i s assumed that a l l parts of the f i g u r e are wider than GS and thus the algorithm should encounter every point within GS of the edge of the f i g u r e . At each white point (x.,y.) along the trace, an area operator, based on equation ( l . l ) , was formed - a quantized disc, 31 g r i d spacings i n diameter. Each point i n s i d e the area operator i s interrogated and a l l those i s found. which are white,are summed. Thus A ' 1 J J where P the white point i n s i d e the area operator. Now A^ i s known to be 725 - ca l c u l a t e d from the known s i z e of the d i s c . A n a l y t i c a l l y A„ = £ P 4 (3.9) 19 th where = the i point i n the area operator. Therefore the ridge function at (x.,y.) i s V , where r 1 r wt-(3.10) For each white point within one g r i d spacing of the nuclear boundary V i s c a l c u l a t e d and stored i n core memory; the discrete-valued ridge function r f o r the nucleus' of the neutrophil i n Figure 3.1, i s shown i n Figure 3.2 . contour-trace ridge f u n c t i o n Figure 3.2 The contour-path and ridge f u n c t i o n of a neutrophil nucleus. The s i z e of the area operator and contour trace can be changed, with respect to the neutrophil image, by varying GS f o r each. "RID" allows the programmer to vary GS from 1 up to 15 fundamental g r i d points of the image. 3.3 Problems Encountered i n Computing the Ridge Function Since the ridge function was implemented on data c o n s i s t i n g of a c t u a l leukocyte images, problems i n forming the area operator and c a l c u l a t i n g the ridge function arose, which would not occur i f t e s t patterns such as the one i n Figure 1.4 were used. I t was the purpose of t h i s thesis, though, to d i s -cover problems a r i s i n g from working with microscopic images so that an eva-l u a t i o n of the usefulness of the v i s u a l - i n p u t device, i n future work, could be made. Id e a l l y , the contour-trace g r i d spacing should be kept as small as possible to minimize quantization noise caused when arc length i s segmented. In a c t u a l p r a c t i c e , when the g r i d spacing was kept small, the contour trace tended to follow the small i r r e g u l a r i t i e s i n the boundary of the neutrophil nucleus - t h i s caused the trace to get caught i n loops and, i n general, produced a poor boundary trace. Since the magnification of the neutrophils i n the trans-parencies i s approaching the l i m i t of o p t i c a l r e s o l u t i o n , the boundary of the c e l l nucleus i s bound to be ragged and "grainy". As the g r i d spacing was increased to above GS=8, the trace d i d not follow the i r r e g u l a r i t i e s and appeared to follow the boundary i n a more continuous manner. (See Figure 3-3). GS=4 GS=8 GS=12 Figure 3-3 Contour-traces showing the e f f e c t s of i n c r e a s i n g the g r i d spacing. As the g r i d spacing was increased above GS=2, the f i x e d s i z e aperture of the image d i s s e c t o r gave a poor average of the transmittance i n the element of area which the g r i d point defines. For instance, i f GS-£, where 3 £ $ ^ 15, the aperture w i l l step to every fundamental image g r i d point and the 1 2 (2 2 mil i n t e r r o g a t i o n area, represents a s p a t i a l area of o mil . This e f f e c t leads to errors i n the contour-trace. (2 2 By simulating a la r g e r aperture which covers the e n t i r e o mil. area, t h i s problem was overcome. The video s i g n a l voltage represents an average of the o p t i c a l density over the area of the e x i s t i n g aperture. Then, to simulate a l a r g e r aperture area, a l l that i s required i s to extend the averaging process over a la r g e r area. A square aperture with a v a r i a b l e side i was simulated. The PDP-9 interrogated a l l fundamental g r i d points w i t h i n the area S and averaged the o p t i c a l density codes i n i t s CPU. The r e s u l t a n t o p t i c a l density code i s a close approximation to that of an actual square aperture with an area o . In a c o l o r l e s s transparency of the actual neutrophil, i t was found that the boundary between the nucleus and the cytoplasm was, at times, poorly defined, e s p e c i a l l y when the cytoplasm contained dark granules. The lack of gre y - l e v e l contrast between the two parts of the leukocyte complicated t h e i r d i s t i n c t i o n . To overcome t h i s problem, a logarithmic a m p l i f i e r was used to expand the contrast between the two regions. Since the A / D analog range, as well as the dynamic range of the video s i g n a l v a r i e d from 0 to -5 v o l t s , a logarithmic a m p l i f i e r with the fo l l o w i n g c h a r a c t e r i s t i c s had to be b u i l t : y ( t ) = m(-x(t)+i) (3.11) where y ( t ) = the output of the non-linear a m p l i f i e r , and x ( t ) = the input video s i g n a l to the a m p l i f i e r . 22 A schematic diagram of the logarithmic a m p l i f i e r i s shown i n Appendix I I I . The transformation from a l i n e a r to a logarithmic o p t i c a l density-scale i s de s i r a b l e since human brightness perception i s also logarithmic and the PDP-9 then works with the same o p t i c a l d e n s i t i e s that a human observer would see. According to the d e f i n i t i o n of the area operator i n Chapter 1, i t i s required that, at each point on the curve C, the disc should enclose only one continuous curvature segment.' Thus, the dis c should not overlap any other portion of the nucleus while operating on a p a r t i c u l a r curvature segment, as shown i n Figure 3.4. The area operator should also have a radius smaller than the thinnest part of the f i g u r e i t i s processing. Both the above conditions were, at times, unavoidably v i o l a t e d because of the nature of the shape of the actu a l neutrophils processed. area operator radius i s too large nuclear o u t l i n e area operator encloses two segments of the nucleus at one time Figure 3.4 Ty p i c a l errors made when operating on actu a l neutrophil n u c l e i . As the size of the area operator was decreased, to reduce the occurrences of the above mentioned errors, the operator became too s e n s i t i v e 23 to curvature. Quantization noise was also added to the ridge function since the s i z e of the contour-trace steps i s increased with respect to the area operator s i z e . Figure 3.5 shows the e f f e c t s of increased ridge function s e n s i t i v i t y to curvature, and increased quantization noise. area operator GS=8 area operator GS=5 Figure 3.5 Ridge functions derived using d i f f e r e n t area operator s i z e s . 4. A NEUTROPHIL CLASSIFICATION STUDY Having the means to produce a ridge f u n c t i o n which i s a shape descriptor, i t i s then possible to a c t u a l l y t e s t i f the f u n c t i o n can be used f o r leukocyte c l a s s i f i c a t i o n . The f o l l o w i n g problem was posed: to determine the age and l o b u l a r i t y of neutrophils. Of the f i v e major morphological types of leukocytes, neutrophils were chosen as a test c l a s s because they seem to have a f a i r l y well defined boundary between the nucleus and cyto-plasm. The morphological ages of i n t e r e s t f a l l i n t o three basic groups: the metamyelocyte, the banded and the segmented c e l l s . Banded and segmented c e l l s are abundant i n healthy peripheral blood, while metamyelocytes are r a r e l y 24 f o u n d . The t y p i c a l n u c l e a r shape o f each o f t h e t h r e e g r o u p s i s shown i n F i g u r e 4.1. F i g u r e 4.1 The g e n e r a l n u c l e a r shapes o f n e u t r o p h i l s . A l t h o u g h a g i n g i s a c o n t i n u o u s p r o c e s s , f o r q u a n t i t a t i v e d e c i s i o n s t o made, s h a r p d e c i s i o n "boundaries must be drawn. The m e t a m y e l o c y t e n u c l e u s i s k i d n e y o r bean shaped; the ba:id c e l l n u c l e u s i s h o r s e - s h o e shaped and o c c u p i e s a s m a l l e r a r e a t h a n t h a t o f t h e m e t a m y e l o c y t e ; and, the segmented n u c l e u s has l o b e s o r i s l a n d s i n t e r c o n n e c t e d by f i l a m e n t s . (These d e f i n i t i o n s a r e p a r a -phased f r o m a he m a t o l o g y manual [ l 4 ] ) . The r i d g e f u n c t i o n f o r m e t a m y e l o c y t e s s h o u l d be r e l a t i v e l y smooth w i t h a s m a l l hump; t h a t , f o r t h e band n u c l e u s s h o u l d have one main peak c o r r e s p o n d i n g ' t o t h e concave p a r t o f t h e h o r s e s h o e ; t h a t , f o r t h e segmented n u c l e u s s h o u l d be i r r e g u l a r and have more t h a n one peak. F i g u r e 4.2 shows a t y p i c a l r i d g e f u n c t i o n f o r e a c h o f t h e age g r o u p s . 25 metamyelocyte banded segmented Figure 4.2 T y p i c a l ridge functions. A peak of curvature i s defined, here, to be above the threshold l i n e . Two peaks cannot l i e adjacent with a spacing l e s s than one f i f t h the length of the ridge f u n c t i o n . (This d e f i n i t i o n of a peak t r i e s to exclude quantization noise and nuclear i r r e g u l a r i t i e s which are not dominant). Looking at Figure 4.1, i t may be noticed that a property of aging i s the formation of lobes i n the nucleus. Generally, the nuclear boundary becomes more i r r e g u l a r as the c e l l ' s age increases. A parameter which gives a measure of the i r r e g u l a r i t y of a f i g u r e boundary i s T/TA] where P = the perimeter of the f i g u r e . A = the area occupied by the f i g u r e . This parameter, which has a lower bound of 24% i n the case of a perfect c i r c l e , i s s i z e and o r i e n t a t i o n independent. P was found by counting the number of g r i d points i n the contour-trace, and the area of the nucleus, A, was found by counting a l l black g r i d points. Thus the feature can be extracted q u a n t i t a t i v e l y on the PDP-9. T h i r t y transparencies of c a r e f u l l y chosen neutrophils, ten of each of the three age groups were processed on the PDP-9. The age of each c e l l was determined by a q u a l i f i e d s p e c i a l i s t [ l 5 J . Figure 4.3 shows a two-dimensional p l o t i n measurement space; the two features used to c l a s s i f y neutrophils are the number of peaks per nucleus and P/VT. Figure 4.3 shows that the three c e l l age groups are d i s t i n g u i s h a b l e i n measurement space - there i s d e f i n i t e c l u s t e r i n g which allows the implemen-t a t i o n of d e c i s i o n boundaries. I t i s then reasonable to assume that the shape of the nucleus, pro-cessed from the ridge f u n c t i o n and the parameter P/7A~is a means of determining the age of neutrophils. A function of curvature with respect to arc length was found u s e f u l as a d e s c r i p t o r of the shape of n u c l e i and i n turn t h e i r c l a s s i f i c a t i o n , but the use of the area operator to derive t h i s f u n c t i o n has four d i s t i n c t d i s -advantages: a) to compute one point on the ridge function, 725 g r i d points which make up the area operator must be interrogated - therefore the process i s i n h e r e n t l y slow. b) the area operator overlapped the nucleus at t h i n regions g i v i n g errors i n the ridge function. c) the area operator impinged on more than one segment of the nucleus at one time, e s p e c i a l l y with segmented n u c l e i , g i v i n g another form of e r r o r i n the ridge function. 27 10 P - S M METAMYELOCYTE B BANDED CELL S SEGMENTED CELL s 8 B B B B B B B M M M MM M MM S S s B B S S 5 S S J L J I 0 1 2 3 4 5 NUMBER OF PEAKS OF RIDGE FUNCTION Figure 4.3 A measurement-space graph showing the c l u s t e r i n g of neutrophil age groups. d) A l i n e a r reading of curvature was not made i n most cases since the radius of curvature of the nucleus was r a r e l y l a r g e r than the radius of the area operator. The statements b) and c) above, have been i l l u s t r a t e d i n Figure 3-4. For the general pattern c l a s s of leukocytes, the area operator provides a poor method of e x t r a c t i n g nuclear curvature. At small operator r a d i i , where the errors o u t l i n e d i n b) and c) are not encountered, the quantization noise l e v e l i n the ridge f u n c t i o n becomes too high. When the s i z e of the operator i s increased, the required curvature s e n s i t i v i t y i s decreased and the ridge f u n c t i o n becomes erroneous. Then,.to overcome the disadvantages mentioned above, an approximation method of curvature measurement, as used by Ledley, should be implemented. The approximation method required information, only from the contour i t s e l f , not from an area around the contour, because i t i s based on the d e f i n i t i o n of curvature dt K = (4.1) ds where dt = the d i f f e r e n t i a l change i n tangent vector on the curve C. ds = the d i f f e r e n t i a l change i n arc length along C. Assume that the curve C i s quantized into equal length segments at points s^, Sg, s ^ , ...s^,..., then the curvature at a point of arc length s^ can be approximated by the expression: s. ) - ( s . - s .) (4.2) K | l( s k - S i ) - S . i } s • s. - s . 1 I 1 J where s, = s. + nAs and s . = s. - nAs. k x j x and As = basic arc length quantization ( i . e . As=s^ +^-s^). n = a non-zero, p o s i t i v e integer. 29 The curvature, according to equation 4.2, can r e a d i l y be c a l c u l a t e d using the contour-trace as a segmented curve. This algorithm would produce a function g i v i n g curvature with respect to arc length and i s obviously much l e s s time consuming than the formation of an area operator. 5. THE SYSTEM PARAMETERS 5.1 Noise C h a r a c t e r i s t i c s of the Image Dissector Tube The theory of the image d i s s e c t o r tube, unlike the orthicon or v i d i c o n tubes, i s simple and thus i t i s r e l a t i v e l y easy to formulate basic noise l i m i t a t i o n s . This theory plays an important r o l e i n the design of the v i s u a l - i n p u t device, since the noise l e v e l l i m i t s the d e t e c t a b i l i t y of the s i g n a l and thus the number of l e v e l s i n t o which the video s i g n a l can be quantized f o r d i g i t a l processing. The detection of a s i g n a l i s hampered by the f o l l o w i n g noise sources: a) Residual dark noise or the thermionic emission of current, detected with no f l u x i n c i d e n t on the photocathode, b) Monitoring c i r c u i t dark noise, c) Inherent random f l u c t u a t i o n s of the s i g n a l f l u x , causing s t a t i s -t i c a l v a r i a t i o n s i n the s i g n a l current, and d) Secondary emission noise introduced by the photomultiplier. For t h i s a p p l i c a t i o n of the image d i s s e c t o r tube, the tube dark noise was found to be n e g l i g i b l e , since i t i s at l e a s t f i v e orders of magnitude smaller than the minimum s i g n a l current. (See reference [ l 7 ] ) . The monitoring c i r c u i t or wide-band video a m p l i f i e r introduced low frequency noise i n t o the system. Under a c t u a l working conditions, 60 Hz hum with some t h i r d harmonic d i s t o r t i o n was observed. The peak-to-peak noise voltage superimposed on the video s i g n a l was about .25 v o l t s over a dynamic r a n g e o f 5.5 v o l t s . T h i s n o i s e c a n be r e d u c e d by c a r e f u l p l a c e m e n t o f the power s u p p l i e s and s h i e l d i n g o f the image d i s s e c t o r tube and v i d e o a m p l i f i e r . A n o t c h f i l t e r c o u l d be employed t o r e d u c e t h e n o i s e power p r e s e n t a t 60 Hz. The f u n d a m e n t a l n o i s e , c a u s e d by s t a t i s t i c a l f l u c t u a t i o n s o f t h e s i g n a l c u r r e n t i s s h o t n o i s e , w h i c h i s " w h i t e " . The mean sq u a r e n o i s e c u r r e n t i s g i v e n by i 2 = 2 q I k A f (5.1) where . q = t h e c h a r g e on an e l e c t r o n . I = t h e d.c. s i g n a l c u r r e n t e n t e r i n g t h e a p e r t u r e . A f = t h e b a n d w i d t h o f t h e v i d e o s i g n a l . The e f f e c t o f t h i s f u n d a m e n t a l n o i s e i n c o n j u n c t i o n w i t h s e c o n d a r y e m i s s i o n n o i s e g i v e s t h e f o l l o w i n g p e a k - t o - p e a k s i g n a l - t o - n o i s e c u r r e n t r a t i o , d e r i v e d i n [ l 8 ] : S N , J A ( c r-i) 1 ' k (5.2) 7 2CTq Af P-P •' where J = t h e d.c. c u r r e n t d e n s i t y i m p i n g i n g on t h e a p e r t u r e p l a n e . A = t h e a r e a o f the a p e r t u r e . d- t h e a v e r a g e g a i n p e r s t a g e o f t h e p h o t o m u l t i p l i e r . As e q u a t i o n (5«l) p o i n t s o u t , t h e n o i s e a m p l i t u d e i s l a r g e s t a t t h e l a r g e s t v a l u e o f I and t h e number o f s i g n a l q u a n t i z a t i o n l e v e l s t h a t c a n be .K. u s e d must be c a l c u l a t e d a t t h i s c o n d i t i o n . O t h e r t h a n c h a n g i n g t h e i n t e r n a l d e s i g n o f t h e image d i s s e c t o r t u b e , the o n l y way t o r e d u c e t h e n o i s e l e v e l i s to d e c r e a s e t h e b a n d w i d t h , A f , o f the e x t e r n a l l o w - p a s s f i l t e r . S i n c e . A f can be t h e o r e t i c a l l y made a r b i t r a r i l y s m a l l , t h e A.&fo m o n i t o r i n g c i r c u i t d a r k n o i s e p r o v e d t o be t h e l i m i t i n g f a c t o r f o r n o i s e . Then, t h e v i d e o s i g n a l can be m e a n i n g f u l l y q u a n t i z e d t o , a t most, e i g h t l e v e l s o r t h r e e b i n a r y b i t s . T h i s was adequate g r e y - l e v e l r e s o l u t i o n f o r preliminary research work. More than eight l e v e l s can be resolved by reducing the monitoring c i r c u i t noise and decreasing the low-pass f i l t e r bandwidth. 5.2 The S p a t i a l Frequency Spectrum Before the bandwidth of the video s i g n a l can be found, i t i s necessary to consider the s p a t i a l frequencies which.exist on the o p t i c a l image. A s p a t i a l frequency can be defined as the rate of change of the o p t i c a l density i n the two-dimensional image plane. The image magnification f i x e s the s p a t i a l frequencies and they are s a i d to e x i s t i n the s p a t i a l domain. When the o p t i c a l density along some curve i n the two-dimensional image i s scanned by the aperture of the image "dissector tube, a time varying video f u n c t i o n i s produced which i s proportional to the o p t i c a l density. The scanning process i s s a i d to transform the s p a t i a l v a r i a t i o n s i n t o the time domain. Assume the s p a t i a l o p t i c a l density v a r i a t i o n s on the image, are given by the two-variable f u n c t i o n f (x,y). Consider a scan along an a r b i t r a r y curve, X" ; the x and y coordinates can be expressed i n terms of the s i n g l e v a r i a b l e arc length along Y. That i s , f'(x,y) = f(s) (5.3) where s i s arc length measured along the curve Y. The aperture a c t u a l l y scans the image i n a d i g i t a l manner, using quantized increments As. Now As = rAt (5.4) where As = s p a t i a l , arc length increments - f i x e d at about 1 mil i n t h i s case, r = the rate of scan along X. At = the ''dwell time" or the time required by the aperture to interrogate a point. Since As i s very small compared to the .8 inch by .8 inch expanse of the image, equation (5.4) can be approximated, with l i t t l e error, by s = r t (5.5) Using equation ( 5 - 5 ) , the time domain representation of f ( s ) can be stated as v ( t ) = f ( J ) (5.6) v ( t ) , then, i s the time domain representation of the function f ( s ) and i s the video s i g n a l produced by a scan along The F o u r i e r transform of v ( t ) i s defined to be V ( f ^ ) . That i s , v ( t ) — V ( f t ) ^ F { v ( t ) ] (5.7) A s i m i l a r d e f i n i t i o n can be applied to f ( s ) , f ( s ) — F ( f s ) ^ F { f ( s ) } (5.8) Now, since v ( t ) = f ( f ) (5,9) and r has some non-zero, p o s i t i v e value f ( J ) — r F ( r . f ) (5.10) by the s c a l i n g property of F o u r i e r transforms. Therefore V ( f t ) = r F(r.f g) (5.11) Equation ( 5 . 1 l ) c l e a r l y shows that the time-domain Four i e r spectrum of f ( s ) , V ( f ^ ) , i s frequency scaled by the rate constant r. By decreasing r, V(f^_) can be compressed. The a b i l i t y to frequency-compress the F o u r i e r spec-trum of the video s i g n a l v ( t ) i s important, since t h i s allows a decrease i n the external bandwidth. Then, the fundamental noise power can be reduced. The neutrophil image, present on the photocathode, represents an o p t i c a l magnification of about 1400 from the actual c e l l . At t h i s magnifica-t i o n , the image contains information approaching the l i m i t of o p t i c a l r e s o l u -t i o n . This l i m i t i s dependent on the wavelength of the l i g h t i l l u m i n a t i n g the c e l l and the imperfections of the magnifying lenses. T h e o r e t i c a l l y , a point source of l i g h t i s transformed i n t o an A i r y d i f f r a c t i o n pattern [20]. The l i m i t of r e s o l u t i o n i s u s u a l l y defined to be twice the distance from the maxima to the f i r s t minima of the A i r y pattern. A n a l y t i c a l l y , t h i s l i m i t i s expressed i n the fo l l o w i n g equation: where R = the r e s o l u t i o n l i m i t distance (expressed i n the • same u n i t s as A.). X - the wavelength of the i l l u m i n a t i n g l i g h t . N.A. = the numerical aperture of the microscope objective . For green l i g h t , the r e s o l u t i o n l i m i t i s about .25_/t. A neutrophil has an average diameter of about 12^ and i t s magnified image on the photocathode w i l l have a diameter of 16.8 mm ( m i l l i m e t e r s ) . The • 25yit r e s o l u t i o n l i m i t i s magnified to .35 mm. This implies that no meaningful o p t i c a l density information i s present at distances smaller than .35 mm on the image. I t i s then evident that s p a t i a l o p t i c a l density frequencies have an upper bound. The maximum s p a t i a l frequency of F ( f g ) , f s _ m a x > c a n then be defined as f =7 (5.13) s-max R w Since F ( f ) i s band-limited to f , V(f,) w i l l also have an upper s s-max t bound frequency of f, , where t-max 34 f\ = r f (5.14) t-max s-max 5.3 The S p a t i a l Resolution For leukocyte c l a s s i f i c a t i o n , the image d i s s e c t o r tube should be required to resolve the distance between two l i n e s , black and white, at a depth of modulation of 100$. In other words, an o p t i c a l density s i g n a l t r a v e r s i n g the complete dynamic range within two fundamental g r i d points should be detectable. Using t h i s d e f i n i t i o n , three d i f f e r e n t forms of s p a t i a l r e s o l u t i o n can be formulated. F i r s t l y , since the computer v i s u a l - i n p u t system i s designed f o r a s p e c i f i c use, operating on a r e s t r i c t e d c l a s s of images, the s p a t i a l frequency spectrum of the images may be formulated. Then, a minimum s p a t i a l r e s o l u t i o n can be defined. I f P(f ) i s found to have an upper bound of f , then the s p a t i a l x s' s-max si g n a l r e s o l u t i o n , R , i s to sp R =-7 (5.15) sp f s-max In t h i s p a r t i c u l a r case, the s p a t i a l s i g n a l r e s o l u t i o n i s simply the o p t i c a l r e s o l u t i o n l i m i t which i s 14 mils, on the image. Secondly, the degradation of the image by the device can be modelled as shown i n Figure 5-1 input picture e*— u(x,y) h'(x,y) o p t i c a l image focus o p t i c a l l y focused image • (x,y) g(x,y) magnetic image focus output image ®> r(x,y) Figure 5.1 A model of the degradation of an image. 35 Assume the input picture i s a step function i n o p t i c a l density, going from black to white i n the x d i r e c t i o n . Then the input i s of the form u(x,y) = u C x , ^ ) (5.16) The lenses focus u(x,y) onto the photocathode g i v i n g r i s e to an A i r y d i f f r a c t i o n pattern. The o p t i c a l image undergoes a transformation, given by the impulse response h (x,y), which i s of the form s i n k x 0 s i n k y „ h(x,y) „ k 0 ( — ^ - ) 2 ( — ± - ) 2 (5.17) 2 y k y The o p t i c a l l y focused image i s converted i n t o an e l e c t r o n i c image by the photocathode. Sharp edges of the e l e c t r o n i c image are defocused i n a Gaussian manner because of the p r o b a b i l i s t i c nature of ele c t r o n emission from the photocathode. Magnetic focusing can be modelled by the s p a t i a l impulse response g(x,y) = k 5 e ~ V 2 - V 2 (5.18) I t i s formulated that an o p t i c a l step function i s defocused into an edge, transformed i n general form according equations (5.17) and (5.18); i t s actual extent (the values of the constants) can only be found experimentally. The edge of a razor blade was focused as best possible, f i r s t o p t i c a l l y onto the photocathode and then magnetically onto the aperture plane, to form an o p t i c a l density step f u n c t i o n . Figure 5.2 i s the quantized o p t i c a l density step response. Each l e v e l corresponds to the o p t i c a l density of an i n t e r -rogation of a fundamental g r i d point. A minimum black to white, t r a n s i t i o n of four fundamental g r i d points, or about 4 mils on the image, can be seen. 36 ^-edge of razor enlarged scan of black-white t r a n s i t i o n Figure 5.2 A scan across the edge of a razor blade. T h i r d l y , when the v i s u a l - i n p u t device i s used under PDP-9 c o n t r o l , the minimum s p a t i a l r e s o l u t i o n which can be detected i s the distance between two fundamental g r i d points. I t i s then obvious that the computer cannot detect a s i g n a l , at a depth of modulation of 100$, whose o p t i c a l density-v a r i e s from black to white i n a distance l e s s than about 2 mils, on the image. 5.4 Bandwidth Requirements The frequency spectrum of the video s i g n a l emerging from the image dis s e c t o r tube c o n s i s t s b a s i c a l l y of the sum of the s i g n a l spectrum V(f^) and a constant amplitude white noise spectrum. Low-pass f i l t e r i n g at the optimal bandwidth should eliminate as much noise power without a s u b s t a n t i a l loss of s i g n a l power. In the d i s p l a y mode, the image i s scanned by adjacent steps of the aperture. Since the bandwidth of the device should be made as small as possible to reduce the noise power, Af i s made equal to f, . This band-37 width i s sub-optimal but adequate, since the exact form of "V(f^) i s not known. I f the detection of the highest frequency possible i s required, then Af should be wide enough so that each aperture i n t e r r o g a t i o n i s barely "seen", or i n other words, so that only the f i r s t harmonic of the s i g n a l i s not f i l t e r e d out. Then, according to the sampling theorem A f = 2 A T (5.19) For the computer-read mode, the external bandwidth should be as small as possible to reduce the noise but not so small that the inherent delay caused by the low-pass f i l t e r w i l l be so long that the f e t c h cycle i s excessively time-consuming. Since a f i n i t e time i s required f o r the d e f l e c t i o n a m p l i f i e r s to s e t t l e and since a s i g n a l propagation delay i s inherent to a low-pass f i l t e r , the A / D conversion of the video s i g n a l should s t a r t a f t e r the s i g n a l reaches a d.c. l e v e l to within the analog voltage equivalent of one-half the le a s t s i g n i f i c a n t b i t of quantization. At the maximum d.c. l e v e l of the video s i g n a l , Af was extended to the point where the peak-to-peak signal-to-noise r a t i o was no smaller than 16:1. This s e t t i n g provides the l e a s t f i l t e r delay possible, before A / D conversion begins, because i t i s the widest bandwidth allowing three b i t g r e y - l e v e l r e s o l u t i o n . 38 6. CONCLUSIONS The v i s u a l - i n p u t system i s a general purpose three-dimensional gra p h i c a l input to the PDP-9. The system was designed to be adaptable to a PDP-9 computer f o r the purpose of eventually automating leukocyte c l a s s i f i c a t i o n but i t can be used f o r any form of computerized image processing. I t can be applied equally well to b i o l o g i c a l image processing, character recognition, picture processing research or a r t i f i c i a l i n t e l l i g e n c e studies. The system can meaningfully resolve eight l e v e l s of o p t i c a l density, u s i n g e i t h e r a l i n e a r or logarithmic scale. A picture matrix of 1024 by 1024 g r i d points can be randomly accessed. O p t i c a l d e n s i t i e s are read i n t o the PDP-9 with an i n t e r r o g a t i o n time dependent on the low-pass f i l t e r bandwidth. (An i n t e r r o g a t i o n time of 100 microseconds i s t y p i c a l ) . A picture i s processed from a 23 mm by 23 mm square on a transparent s l i d e ; the image f i e l d can be viewed on an e x t e r n a l l y c o n t r o l l e d d i s p l a y u n i t . The programmer has the option of on-line processing or sys t e m a t i c a l l y reading the picture information onto magnetic tape (D.E.C. tape) and proceeding with data processing from that medium. The image d i s s e c t o r tube aperture, which was 1 mil i n diameter, was found to be needlessly small. I t i s a major f a c t o r c o n t r i b u t i n g to a small low-pass f i l t e r bandwidth and thus' a long g r i d point i n t e r r o g a t i o n time, and i t provided unnecessary s p a t i a l r e s o l u t i o n , below an inherent focusing l i m i t of 4 mils. The simulation of l a r g e r apertures showed that an aperture l a r g e r than 4 mils proved to be a better picture transducer, reducing j i t t e r . The successful c l u s t e r i n g , i n measurement space, of the neutrophil n u c l e i , strongly i n d i c a t e s that c l a s s i f i c a t i o n of leukocytes using curvature functions should be f u r t h e r i n v e s t i g a t e d . Instead of Connor's area operator, though, a curvature e x t r a c t i o n method of the type used by Ledley should be implemented. More important, the neutrophil c l a s s i f i c a t i o n study has shown that the v i s u a l - i n p u t system can be s u c c e s s f u l l y used f o r the morphological c l a s s i f i c a t i o n of leukocytes, making f e a s i b l e the automation of the d i f f e r e n t i a l blood count. Future leukocyte c l a s s i f i c a t i o n work should incorporate d e s c r i p t i v e features such as color, nuclear density, the r e l a t i v e frequency of c e l l types, and g r a n u l a r i t y of the cytoplasm, as well as morphological shape. Problems may be encountered i n the detection of f o l d s i n n u c l e i and also the d i s t i n c t i o n of n u c l e i themselves where there i s a lack of g r e y - l e v e l contrast between the nucleus and cytoplasm. Inconsistent c e l l s t a i n i n g w i l l cause varying grey-l e v e l ranges f o r c e l l s of the same type, and the i l l u m i n a t i o n w i l l have to be adjusted. 40 APPENDIX I S p e c i f i c a t i o n s and. Performance of a Five m i l Aperture Image Dissector Tube Since the monitoring c i r c u i t dark noise can be reduced by c a r e f u l design, assume, f o r t h i s discussion, that the l i m i t i n g noise f a c t o r i s fun-damental noise, with a signal-to-noise r a t i o given by equation (5.2). During contour-tracing and preprocessing, the e x i s t i n g 1 mil diameter aperture proved to be needlessly small. I t was smaller than the r e s o l u t i o n l i m i t of the focusing system of the image d i s s e c t o r tube. Only three b i t s of g r e y - l e v e l quantization were used and thus a signal-to-noise r a t i o s l i g h t l y smaller than 16:1, at maximum I , i s required. The low-pass bandwidth was adjusted accordingly. Then, to improve both the signal-to-noise r a t i o and s p a t i a l r e s o l u -t i o n , a l a r g e r aperture tube should be purchased. An aperture diameter of 5 mils or .127 mm i s appropriate. The aperture diameter w i l l then be j u s t l a r g e r than the "best-focus" of an edge and s i g n a l s with a depth of modulation of 100^  between two funda-mental g r i d points, spaced at about 5 mil i n t e r v a l s , w i l l be e a s i l y detected. 2 Because the new aperture would average the o p t i c a l density over a 19.5 m i l area, j i t t e r w i l l be l e s s noticeable. The o p t i c a l r e s o l u t i o n l i m i t of actual leukocytes w i l l be enlarged to 5 mils at an image magnification of about 500. G r i d point i n t e r r o g a t i o n at t h i s magnification w i l l r e s u l t i n no loss of meaningful s p a t i a l r e s o l u t i o n . (The image emerging from the eyepiece of a 500 power microscope can then be projected d i r e c t l y onto the photocathode of the image d i s s e c t o r tube). With a 5 mil aperture, the signal-to-noise r a t i o w i l l be increased by a f a c t o r of 19«5> i f the external bandwidth remains the same as above. Then a signal-to-noise r a t i o of 64-'. 1 can be achieved and 6 b i t s of g r e y - l e v e l quantization are meaningful. A l t e r n a t e l y , maintaining 3 b i t s of g r e y - l e v e l r e s o l u t i o n , the external bandwidth can be extended by a f a c t o r of 19.5. This w i l l r e s u l t i n a smaller low-pass f i l t e r delay time and image processing w i l l be f a s t e r . APPENDIX II D e t a i l s of the Hardware Layout Figure 2.1 shows a block diagram of the layout of the i n t e r f a c e between the image d i s s e c t o r camera and the PDP-9, and the hardware required to- produce a dis p l a y . The l o g i c a l functions were wired using D.E.C. plug-in modules. The d e t a i l s of the l o g i c networks which are contained i n each block of Figure 2.1, are given i n t h i s appendix. PiTL LOGIC 0V--+3V LOGIC P O W E R C L E A R B » — D E V I C E S E L E C T I O N L I N E S S U B D E V I C E S E L E C T I O N - | L I N E S I O P L I N E S E N A B L E D A T A I / O S K I P R Q (IOT1 D O ) S T O P C L O C K L O A D X ( N O T U S E D ) (IOT1 D2) S T O P C L O C K L O A D Y ( N O T U S E D ) S T A R T A / D C O N V . S T A R T C L O C K C L E A R S Y S T E M 1 (_MAN. 1 C L E A R R E A D A / D C L E A R A / D F L A G A / D C O N V . C O M P L E T E Figure A2.1 Device c o n t r o l . (I0T1 DO) STOP CLOCK (IOT1 D2) STOP CLOCK START CLOCK MAN. A STOP H CLOCK 4" MAN. I START H CLOCK 1 D/A INPUT CONTROL CLOCK ENABLE SPEED ' SET VARIABLE SPEED CLOCK CLOCK OUTPUT Figure A2.2 Mode c o n t r o l . 8 1 0 11 12 1 3 1 4 1 5 1 6 1 7 i> 4> D 7 0 <3^  •o—:=;L> O f > = 3 > READ REQUEST S L O O S L 0 2 .-—.ex 1 8 Y .9X - 9 Y 1 0 X 1 0 Y . 1 1 X • 1 1 Y 1 2 X 1 2 Y ^ 1 3 X I 1 3 Y . 1 4 X - 1 4 Y . 1 5 X - 1 5 Y . 1 6 X - 1 6 Y 1 7 X 1 7 Y T O D / A ' C O N V E R T E R S DL12 DL13 DL14 DL15 DL16 DL17 I / O B U S DATA L I N E S FROM A/D CONVERTER R E A D A / D 0 —X 1 —X 2 ~H>— —K 3 -0- —X 4 —X 5 H>-—X 6 -t>-—X 7 —K N O T ' U S E D Figure A 2 . 3 Data t r a n s f e r c o n t r o l . U l C L E A R A / D F L A G C L E A R S Y S T E M A / D C O O T . C O M P . L E V E L C O N V E R T E R D E C - R T L s 1 R 0 L E V E L C O N V E R T E R D E C - R T L R T L C L O C K S T A R T A / D C O N V E R S I O N s 1 R 0 L E V E L C O N V E R T E R R T L - D E C A / D D O N E C O M P A R A T O R V I D E O S I G N A L S 1 S 1 C 0 C 0 L E V E L C O N V E R T E R R T L - D E C L E V E L C O N V E R T E R R T L - D E C L E V E L C O N V E R T E R R T L - D E C L E V E L C O N V E R T E R -R T L - D E C L E V E L **i C O N V E R T E R R T L - D E C L E V E L C O N V E R T E R R T L - D E C R-2R L A D D E R Figure A2.4 A / D converter. X R I P P L E C O U N T E R C L E A R S Y S T E M C L O C K O U T P U T s 1 T C 0 S 1 T C 0 IX I n s 1 T C 0 2X s 1 T C 0 I T O D A T A G A T E S Y C O U N T E R T O G G L E \ s 1 T C 0 t MAN. S T O P Y C O U N T E R 7X J Y R I P P L E C O U N T E R C L E A R S Y S T E M Y C O U N T E R T O G G L E JT s 1 T C 0 s 1 T c 0 s 1 T c 0 s 1 T c 0 6Y T O D A T A G A T E S Figure A2.5 8 B i t X (Y) r i p p l e counter. 7Y J X L O A D P U L S E C O N T R O L C L O C K O U T P U T M O N O -S T A B L E P U L S E C O N V E R T E R > C L E A R - S Y S T E M L O A D X L O A D Y P U L S E L E V E L C O N V E R T E R D E C - R T L L O A D X D A T A ? r Y L O A D P U L S E C O N T R O L C O U N T E R T O G G L E M O N O -S T A B L E L O A D Y P U L S E C L E A R S Y S T E M L O A D Y L E V E L C O N V E R T E R D E C - R T L L O A D Y D A T A Figure A 2 . 6 X (Y) load pulse c o n t r o l . x D/A OUTPUT (Y D/A OUTPUT) 3-9K > 3.9K 2.2K 2.2K I DISPLAY X DEFLECTION (DISPLAY Y DEFLECTION) CAMERA X DEFLECTION (CAMERA Y DEFLECTION) F i g u r e A2.7 X (Y) a t t e n u a t o r . F R O M i / O D A T A L I N E S F R O M D A T A • G A T E S 1 7 X ( 1 7 Y ) 1 6 X ( 1 6 Y ) 1 5 X ( 1 5 Y ) 1 4 X ( 1 4 Y ) 13X (13Y) 1 2 X ( 1 2 Y ) 1 1 X ( l l Y ) 1 0 X ( I O Y ) 9X (9Y) 8 X ( 8 Y ) L . C . D E C - R T L L . C . D E C - R T L L . C . D E C - R T L L . C . D E C - R T L " L . C . D E C - R T L L . C . D E C - R T L L . C . D E C - R T L L . C . D E C - R T L L . C . D E C - R T L L O A D X D A T A ( L O A D Y D A T A L . C . D E C - R T L 10 B I T R E G I S T E R B I N A R Y W E I G H T E D L A D D E R X D / A O U T P U T ( Y D / A O U T P U T ) 3»-Figure A 2 . 8 X ( Y ) D / A converter. FROM X RIPPLE COUNTER (FROM Y RIPPLE COUNTER) D / A INPUT CONTROL \ 4X (4Y) 5 X (5Y) 6 X ( 6 Y ) 7 X ( 7 Y ) 0 0 0 0 0 I I 13X (131) 1 2 X ( 1 2 Y ) 1 1 X ( H Y ) 1 0 X ( 1 0 Y ) 9X (9Y) V 8 X ( 8 Y ) y T O x D / A C O N V E R T E R ( T O Y D / A C O N V E R T E R ) Figure A 2 . 9 . X ( Y ) data gates. 52 APPENDIX I I I The Logarithmic A m p l i f i e r A logarithmic a m p l i f i e r was used as an optional feature to provide a logarithmic g r e y - l e v e l s c a l e . A schematic diagram of the a m p l i f i e r i s shown below. ALL AMPLIFIERS NEXUS SQ-lOa Figure A3.1 Schematic of the logarithmic a m p l i f i e r . 53 BIBLIOGRAPHY 1. Stein, P.G., L i p k i n , L.E., .and Shapiro, H.M. "Spectre I I : General-Purpose Microscope Input f o r a Computer" Science, 1966, October 1969, ( 3 2 8 - 3 3 3 ) . 2. .Ledley, R.S. "Automatic Pattern Recognition f o r C l i n i c a l Medicine" Proceeding of the IEEE, 57, 11, November 1969, (2017-2035). 3. Bostrom, R.C. and Holcomb, W.G. "CYDAC—A D i g i t a l Scanning Cytophotometer" IEEE I n t e r n a t i o n a l Convention Records, VII, 1963, ( 1 1 0 - 1 1 9 ) . 4. Mawdesley - Thomas, L.E. and Healey, P. "Automated An a l y s i s of C e l l u l a r Changes i n H i s t o l o g i c a l Sections" Science, 163, March 1969, ( 1 2 0 0 ) . 5. Rosenberg, S.A., Ledeen, K.S., and K l i n e , T. "Automatic I d e n t i f i c a t i o n and Measurement of C e l l s by Computer" Science, 163, March 1969, (1065-1067). 6. Technical Brochure "Computer Eye" Information I n t e r n a t i o n a l Inc. 7. Tretiak, O.J. "Picture Processing" M.I.T. Quarterly Progress Report, Cambridge, No. 83, October 1966, ( 1 2 9 - 1 4 2 ) . 8. Prewitt, J.M.S. and Mendelsohn, M.L. "The Analysis of C e l l Images" , Annals New York Academy of Sciences, 128, 1966, (1035-1053). 9. Young, I.T. " B i o l o g i c a l Image Processing - Automated Leukocyte Recognition" M.I.T. Quarterly Progress Report, Cambridge, No. 89, A p r i l 1968. 10. Bourk, T.R. "Leukocyte Locating Procedure" M.I.T. Quarterly Progress  Report, Cambridge, No. 97, A p r i l 1970. 11. R i t a l a , W.M. and Hsu, C.C. "A Feature-Detection Program f o r Patterns with Overlapping C e l l s " IEEE Transactions, SSC-4, No. 1, March 1968, (16-23). 12. Connor, D.J. " L a t e r a l I n h i b i t i o n and the Area Operator i n V i s u a l Pattern Processing" Ph.D. Thesis, U.B.C., June 1969-13. Mason, S . J . and Clemens, J.K. " C h a r a c t e r R e c o g n i t i o n i n an E x p e r i m e n t a l R e a d i n g Machine f o r t h e B l i n d " R e c o g n i z i n g P a t t e r n s , P.A. K o l e r and M. Eden - ed., The M.I.T. P r e s s , Cambridge Mass., 1968, ( 1 5 6 - 1 6 7 ) . 14. W i n t r o b e , M.M. C l i n i c a l Hematology, L e a and F e b i g e r , P h i l a d e l p h i a , 1951. 15. M o r p h o l o g i c a l V e r i f i c a t i o n o f Data . M i s s D. Johnson, A.R.T., Head o f Hematology Dept., S t . P a u l ' s H o s p i t a l , V a n c o u v e r , B.C., 1970. 16. E b e r h a r d t , E.H. " N o i s e i n M u l t i p l i e r P h o t o t u b e s " R e s e a r c h Memo No. 309, I.T.T. I n d u s t r i a l L a b o r a t o r i e s , I n d i a n a , I 9 6 0 . 17. E b e r h a r d t , E.H. " N o i s e i n Image D i s s e c t o r Tubes" R e s e a r c h Memo No. 337, I.T.T. I n d u s t r i a l L a b o r a t o r i e s , • I n d i a n a , 1961. 18. E b e r h a r d t , E.H. " S i g n a l - t o - N o i s e R a t i o i n Image D i s s e c t o r s " R e s e a r c h Memo No. 386, I.T.T. I n d u s t r i a l L a b o r a t o r i e s , I n d i a n a , 1964. 19- E b e r h a r d t , E.H. " O f f - F o c u s Beam S i z e i n Image D i s s e c t o r s " R e s e a r c h Memo No. 426, I.T.T. I n d u s t r i a l L a b o r a t o r i e s , I n d i a n a , 1965. 20. L o n g h u r s t , R.S. G e o m e t r i a l and P h y s i c a l O p t i c s . Longmans, Green and Co. L t d . , London, 1964. 

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0102199/manifest

Comment

Related Items