Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Proposed protocol for internal dosimetry using patient-specific attenuation-corrected spect scans Hannis, Leah 1999

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

Item Metadata

Download

Media
831-ubc_1999-0195.pdf [ 5.57MB ]
Metadata
JSON: 831-1.0085101.json
JSON-LD: 831-1.0085101-ld.json
RDF/XML (Pretty): 831-1.0085101-rdf.xml
RDF/JSON: 831-1.0085101-rdf.json
Turtle: 831-1.0085101-turtle.txt
N-Triples: 831-1.0085101-rdf-ntriples.txt
Original Record: 831-1.0085101-source.json
Full Text
831-1.0085101-fulltext.txt
Citation
831-1.0085101.ris

Full Text

P R O P O S E D P R O T O C O L F O R I N T E R N A L D O S I M E T R Y USING PATIENT-SPECIFIC A T T E N U A T I O N - C O R R E C T E D S P E C T SCANS By Leah Hannis B . S c . University of Calgary, 1997  A THESIS S U B M I T T E D IN PARTIAL F U L F I L L M E N T O F THE REQUIREMENTS  FOR THE DEGREE OF  M A S T E R OF  SCIENCE  in THE FACULTY OF GRADUATE DEPARTMENT OF  STUDIES  PHYSICS  We accept this thesis as conforming to the required standard  T H E UNIVERSITY OF BRITISH C O L U M B I A  April 1999 © Leah Hannis, 1999  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of B r i t i s h C o l u m b i a , I agree that the L i b r a r y 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 Physics The University of B r i t i s h C o l u m b i a 6224 A g r i c u l t u r a l R o a d Vancouver, B . C . , C a n a d a V 6 T 1Z1  Date:  ABSTRACT The Medical Internal R a d i a t i o n Dose ( M I R D ) protocol has provided a solid foundation over the years for calculating the internal absorbed dose but as for providing an assessment that is specific to the individual patient's organ/tumor anatomy, it falls short. Various methods have been proposed to overcome the shortcomings of the M I R D protocol, but each has its own limitations. Quantitative S P E C T proves to be the most desirable option for determining dose estimates, yet the prolonged scan times prevent S P E C T from becoming a clinical protocol in dosimetry. Given the time constraints of clinical S P E C T , planar quantitative imaging has proved a popular choice for dosimetry studies, but the resulting dose overestimates may prevent m a x i m u m therapy from being achieved. Proposed here is a protocol that contends to be clinically feasible, patient-specific, and promising i n its results. T h i s protocol combines the benefits of both quantitative planar and S P E C T imaging. B y maintaining the majority of scans as planar yet incorporating the benefits of attenuation corrected S P E C T scans, a more accurate, yet attainable clinical protocol can be achieved. T h e 2 or more S P E C T scans suggested make use of a Gadolinium-153 transmission source so that a simultaneous emission/transmission scan provide a patient-specific, attenuation corrected S P E C T image. The S P E C T data is then used to constrain the planar data, resulting in a more accurate dose estimate than would arise from planar alone. P h a n t o m experiments demonstrate that the errors i n absorbed dose estimates have improved from an average of 159% for planar methods alone to 14% by the addition of a S P E C T constraint.  ii  T A B L E OF C O N T E N T S  Abstract  ii  Table of Contents  iii  List of Figures  vi  List of Tables  viii  Acknowledgements 1  ix  Introduction  1  1.1  A i m of T h i s W o r k  2  1.2  Structure of T h i s Thesis  3  1.3  Ionizing R a d i a t i o n  4  1.3.1  4  1.4  Ionizing Particles  Methods of Interaction of Photons .  6  1.4.1  T h e Photoelectric Effect  6  1.4.2  Rayleigh Scattering  7  1.4.3  C o m p t o n Scattering  7  1.5  Attenuation of Photons  8  1.6  G a m m a Camera  8  1.6.1  Collimator  9  1.6.2  Detector C r y s t a l  11  1.6.3  Photomultiplier Tube A r r a y and Position Logic C i r c u i t s  11  1.6.4  Computer  12  iii  2  3  Internal Dosimetry  13  2.1  The Medical Internal R a d i a t i o n Dose ( M I R D ) P r o t o c o l  14  2.1.1  Physical and Biological Half-Lives  14  2.1.2  Source and Target Organs  15  2.1.3  C a l c u l a t i n g the Absorbed Dose  16  Measurement Techniques  21  3.1  22  Planar Imaging  -  22  3.1.2  S P E C T Imaging  22  3.1.3  Attenuation Effects i n Imaging  22  Current Methods for Dosimetry  24  3.3  Quantitative Planar and S P E C T Imaging  27  3.3.1  Attenuation Correction in Planar Imaging  27  3.3.2  Attenuation Correction i n S P E C T  32  Dosimetry M e t h o d Proposed i n this Work  34  Simulations  37  4.1  M a t l a b Simulations  38  4.1.1  Results  42  S i m S E T Simulations  45  4.2.1  48  4.2  5  3.1.1  3.2  3.4 4  Imaging Methods  Results  Phantom Experiments  50  5.1  Technetium-99m P h a n t o m Experiment # 1  51  5.1.1  Acquisition Protocols  51  5.1.2  Scan Schedule  53  iv  5.2  6  5.1.3  Region of Interest (ROI) Analysis  54  5.1.4  S P E C T Constraint  56  Technetium-99m P h a n t o m Experiment # 2  58  5.2.1  Acquisition Protocols  59  5.2.2  Scan Schedule  59  5.2.3  Region of Interest (ROI) Analysis  60  5.2.4  S P E C T Constraint  64  5.3  Discussion  69  5.4  Other Factors Affecting Quantitation  70  5.4.1  P i x e l Saturation  70  5.4.2  Dead-Time  71  Conclusions  75  6.1  Summary of the Work  -75  6.2  Suggestions for C l i n i c a l Applications  77  Bibliography  79  v  L I S T OF F I G U R E S  1.1  Schematic representation of the G a m m a C a m e r a and its components (not to scale)  1.2  9  Top and Side views of the parallel hole collimator frequently used in N u clear Medicine.  In the Side view it is apparent that 7 rays having an  incidence angle greater than the collimator acceptance angle w i l l be absorbed by the collimator 2.1  10  Concept of source and target regions for internal dosimetry calculations. Source regions (shaded regions) contain activity and may contribute to the absorbed dose i n target regions  3.1  16  Schematic representation of (a)planar and ( b ) S P E C T (2 detector heads) imaging and the projection profiles collected by the detectors. P l a n a r and S P E C T projections are 2-dimensional representations of a 3-dimensional object. T h e ability of S P E C T to rotate around the object collecting multiple 2-D views enables the reconstruction of a 3-D image of the object.  3.2  .  23  T h e anterior and posterior projections from a constant source distribution K (shaded object) at a mean depth m w i t h i n the constant attenuator \x.  28  3.3  Sample T i m e - A c t i v i t y curve as used to determine the cumulated activity.  31  3.4  M L A transmission source and resulting profiles (a)without and (b)with the attenuating object  3.5  33  Energy spectrum of the Gd-153 (transmission) and T c - 9 9 m (emission) sources used in a simultaneous emission/transmission scan  vi  34  4.1  Pixelized representation of the attenuating ellipse used i n M a t l a b simulations  4.2  38  Locations of the activity sources and background activity w i t h i n the M a t lab object  4.3  39  Anterior and posterior attenuating matrices. A c t u a l matrices are 64 x 64. Each pixel has an attenuation weighting defined by its depth w i t h i n the attenuating ellipse as measured from the detector head  4.4  40  Representation of the activity distribution (as viewed i n the central slice) and all 12 slices of the attenuating cylinder  4.5  46!  A c t i v i t y positions as viewed i n the central slice of the attenuating cylinder used i n S i m S E T simulations  47  5.1  P l a n a r imaging of a phantom by two detector heads i n a 180° configuration 52  5.2  S P E C T imaging of a phantom by two detector heads i n a 90° configuration 53  5.3  Heart T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 1 . .  5.4  T h o r a x phantom w i t h two spherical tumor inserts - one underneath the right lung and one on the lateral side of the left lung  55  59  5.5  Heart T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 2 .  .  62  5.6  Tumor 1 T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 2.  63  5.7  Tumor 2 T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 2 .  64  5.8  Results of the D e a d - T i m e investigation for detector 1 of the Siemens e.cam camera  5.9  73  Results of the Dead-Time investigation for detector 2 of the Siemens e.cam and detectors 1 and 2 of the Siemens M u l t i S P E C T 2  vii  74  L I S T OF T A B L E S  4.1  M a t l a b results of attenuation correction using planar techniques w i t h fi = 0.15cm"  4.2  1  ( A c t i v i t y Source Combinations are labeled i n Figure 4.2).  . . .  43  M a t l a b results of attenuation correction using planar techniques w i t h /j, = 0.12cm  - 1  ( A c t i v i t y Source Combinations are labeled i n Figure 4.2).  . . .  44,  4.3  S i m S E T results of attenuation correction using planar techniques  48  4.4  S i m S E T investigation of the effects of different levels of background radiation on attenuation correction in planar techniques  49  5.1  Heart data from planar scans (Experiment # 1)  54  5.2  Heart data from S P E C T scans (Experiment # 1)  57  5.3  Planar data re-normalized using the S P E C T Constraint for the Heart R O I (Experiment # 1)  58  5.4  Heart data from planar scans (Experiment # 2 )  60  5.5  Tumor 1 data from planar scans (Experiment #2)  5.6  T u m o r 2 d a t a from planar scans (Experiment # 2 )  61  5.7  Heart data from S P E C T scans (Experiment # 2 )  65  5.8  P l a n a r d a t a re-normalized using the S P E C T Constraint for the Heart R O I  5.9  .  60  (Experiment # 2 )  66  Tumor 1 data from S P E C T scans (Experiment # 2 )  66  5.10 Planar data re-normalized using the S P E C T Constraint for the Tumor 1 R O I (Experiment # 2 )  67  5.11 Tumor 2 data from S P E C T scans (Experiment # 2 )  68  5.12 Planar data re-normalized using the S P E C T Constraint for the T u m o r 2 R O I (Experiment # 2 )  69 viii  ACKNOWLEDGEMENTS Through all the blood, sweat, and tears, I was not alone i n this endeavor.  I thank  A n n a Celler for making all of this possible, Troy Farncombe for helping me through countless jams, A l e x Mackay for his kindness, Janet Johnson and Janet Measday for believing in me, R o n a l d Harrop for his insight, D o n Lyster for his clinical perspective, the Nuclear Medicine technologists for always accomodating me, and many others that lent me wisdom, support, and humor along the way. M o s t precious to me i n this world are my family and friends, without which I would be firmly entrenched between padded walls. I hope to thank you each individually for my gratitude can hardly be expressed here i n words. Your warmth blankets me from a reality that can be unforgiving. Because of you, I live i n a world where wishes do come true. M a y our giggles echo far into the future.  For my brother M i t c h .  ix  CHAPTER 1 INTRODUCTION  Unlike external radiation sources, such as X - r a y tubes and isotope units, Nuclear Medicine utilizes radiation from w i t h i n the body to perform its diagnostic and therapeutic applications. T h i s internal radiation originates from a radiopharmaceutical that has either been injected, ingested, or inhaled by the patient.  T h e radiopharmaceutical is a phar-  maceutical agent w i t h a radionuclide (or radioactive nuclide) bound to it. T h e unstable atom(s) i n the radionuclide can result i n the emission of alpha particles, beta particles, gamma rays, Auger electrons, and/or low energy x-rays. A l l of these types of radiation are capable of giving rise to ionization when they are absorbed i n matter. A b s o r p t i o n of energy from ionizing radiation may cause damage to living tissue and it is therefore of interest i n Internal Dosimetry to quantify the radiation deposited i n tissues and organs. Since Nuclear Medicine is primarily focused on diagnostic imaging, internal dosimetry has been implemented as a means to calculate internally absorbed doses to assess the potential risks to the patient.  For therapeutic applications, ionizing radiation can  be viewed as beneficial when considering damage to unwanted (ie. malignant) tissue, however, it is necessary to quantitatively assess dose distributions to ensure satisfactory therapy while m i n i m i z i n g dose to critical organs. Radionuclide therapy involves the use  1  Chapter 1. Introduction  2  of larger amounts of radioactivity than i n diagnostic studies, thus greater care must be used when assessing absorbed dose distributions i n the body. The M e d i c a l Internal R a d i a t i o n Dose ( M I R D ) Committee of the Society of Nuclear Medicine has established a formalism to provide a general framework for the dosimetry of administered pharmaceuticals.  This framework, which employs the use of planar  imaging techniques, was initially established to aid in the determination of absorbed dose from diagnostic radiopharmaceuticals. Although this method provides a consistent but simplified approach, the estimates are generally conservative - and i n the case of radiotherapy, applicability of the method tends to be quite limited [1].  1.1  A I M OF THIS  WORK  A major obstacle i n attaining accurate quantitation i n dosimetry is the attenuation of photons i n the patient. T h e availability of Single P h o t o n Emission Computed Tomography ( S P E C T ) , together with transmission scanning, allows a more accurate means of determining dose distributions i n the body as compared to planar imaging techniques. The use of a transmission source in S P E C T makes possible the construction of an attenuation map of the patient and, thus, patient-specific correction for the attenuation of photons passing through the patient - allowing for a more accurate estimation of the activity in the body. Transmission sources can be used i n both S P E C T and planar studies but the fact that S P E C T provides a 3-dimensional way of accounting for activity arising from regions surrounding the source regions of interest, makes it superior to quantitative planar imaging. Quantitative S P E C T provides the volumetric information needed for dosimetry whereas the planar scans cannot. Unfortunately this is where the obstacle of practicality comes i n . Regardless of the  Chapter 1. Introduction  3  improvement quantitative S P E C T can provide, it is just not practical i n internal dosimetry to do S P E C T full body scans every time that a scan is required. T h i s is precisely why quantitative planar methods are still used, even i n the presence of quantitative S P E C T . Presently, even a single full body S P E C T acquisition is very time consuming for regular clinical use, let alone implementing multiple full body S P E C T scans as a protocol in internal dosimetry (as multiple scans are required to calculate the internal radiation dose). T h i s work involves an investigation into utilizing the benefits of regional quantitative S P E C T scans i n internal dosimetry while still allowing the majority of scans to be planar. The hope is that the incorporation of attenuation-corrected  S P E C T scan(s) into the  common planar protocol will provide a useful improvement to internal dose estimates while staying w i t h i n the limits of clinical practicality.  1.2  STRUCTURE  OF THIS THESIS  T h i s introductory chapter builds the necessary foundation for an understanding of radiation relevant to Nuclear Medicine, its method of interaction i n matter, and the means by which radiation is detected and stored for further analysis. Chapter 2 on Internal Dosimetry covers the fundamental basis ( M I R D protocol) by which the absorbed dose is calculated. T w o important factors for determining the absorbed dose are introduced: the M I R D S-factors and the cumulated activity A. T h i s thesis focuses on improving the estimation of the cumulated activity i n order to improve the absorbed dose estimates. Measurement Techniques (Chapter 3) presents the imaging modalities available to acquire the patient images that are needed to perform dosimetry, namely planar and S P E C T imaging, the limitations imposed on imaging by the attenuation of photons and how quantitation may be achieved by correcting for attenuation.  Current methods  Chapter 1. Introduction  4  to perform dosimetry are included in this chapter along w i t h the method proposed in this work. In Chapter 4, Simulations are described which quantify the shortcomings introduced by attenuation correction in planar imaging. T h e intent is to demonstrate why quantitative planar imaging is not sufficient as a complete method.  A n d finally,  Phantom Experiments presents some experimental results of the method proposed in this work and acknowledges two important factors that may affect quantitation in dosimetry, namely pixel saturation and dead-time.  1.3  IONIZING  RADIATION  It is important to understand the basic types of particles that are responsible for ionization in matter. T h e following is a brief summary of the origin of these particles and the way in which each behaves in matter. Only those particles of primary importance to internal dosimetry w i l l be discussed further.  1.3.1  ALPHA  IONIZING  PARTICLES  PARTICLES  Some atomic nuclei decay by the emission of an alpha particle. A n alpha particle consists of two neutrons and two protons that are bound together. T h e alpha particle is a heavy particle and it is because of this, and its +2 charge, that it is highly interactive. These factors keep the alpha particle from travelling very far in any material, making it a nonpenetrating radiation. Even alpha particles with energies as high as 5 M e V have ranges of less than 0.04millimetres in tissue [2]. A l p h a particles lose energy by collisions with atomic electrons, causing ionizations to occur. A l p h a particles are generally the result of nuclear decay in heavy elements. If ingested, these heavy elements have a tendency to deposit themselves i n bone, usually interminably, thus increasing the radiation dose to  Chapter 1. Introduction  5  the bone marrow and i m p a i r i n g their usefulness i n Nuclear Medicine applications.  BETA PARTICLES (ELECTRON OR POSITRON  EMISSION)  Nuclear decay can also result i n the emission of electrons or positrons. Electrons are very light particles (7350 times less massive than an alpha particle) therefore less interactive, giving them a longer range i n material [3]. T h e i r paths through material also tend to be highly random. T h e electrons lose energy by collisions w i t h atomic electrons (resulting in ionizations), by close encounters with an atom which can result i n the excitation of the atom and/or by bremsstrahlung (which results i n the emission of photons).  Since  bremsstrahlung is generally an effect of high energy (MeV. range) electrons, which are far above the energies used i n Nuclear Medicine, it is a process not of interest to this thesis. The range of an electron depends on its energy and the density of the material through which it travels.  Its range tends to be longer than that of an alpha particle yet i t is  still considered a non-penetrating (short-ranged) radiation. A s an example, the electrons produced by Hydrogen-3 and Sulphur-35 (having m a x i m u m electron energies of 18 keV and 167 k e V respectively) have average ranges i n unit-density material of 5.2 pm and 0.32 m m respectively [2].  PHOTONS  The gamma photon, the v i t a l component of Nuclear Medicine, is a result of interactions involving the atomic nucleus. O f the radionuclides used i n Nuclear Medicine, the most abundantly used is Technetium-99m. Its mode of producing gamma rays w i l l be discussed here.  Technetium-99m is i n a metastable state, as denoted by the ' m ' . T h e decay of  a metastable state by the emission of a 7 ray is called an isomeric transition.  A s is  common to excited states, a Technetium-99m decay may also result i n the emission of  Chapter 1. Introduction  6  an orbital electron by a process called internal conversion.  In this case, the nucleus  decays by transferring energy to an orbital electron, resulting i n the ejection of the electron instead of a gamma ray. Whether an electron or gamma ray is emitted is a study i n probabilities but since the ratio of photons to electrons emitted is high, Tc-99m proves to be a definite advantage for studies requiring detection of 7 rays from internally administered radioactivity [4].  1.4  METHODS OF INTERACTION OF PHOTONS  A t energies relevant to Nuclear Medicine (less than 511keV), gamma photons can interact with matter by means of the photoelectric effect, Rayleigh scattering, and C o m p t o n scattering (pair production cannot arise as it is only a factor at energies > 1.022MeV). These interactions can lead to ionizations, thus the emission of secondary electrons. T h e higher a photon's energy is, the greater its penetrating capabilities i n matter.  1.4.1  T H EP H O T O E L E C T R I C  EFFECT  In the photoelectric effect the incident photon is completely absorbed by the atomic electron it collides with. T h i s absorption of energy results i n the emission of an orbital electron, known as a photoelectron.  T h e kinetic energy of the emitted electron is the  difference between the energy of the incident photon and the binding energy of the orbital electron. T h e short-ranged behavior of the electron, as mentioned earlier, results i n the photoelectron depositing its energy close to the site of the photoelectric interaction. B y definition, the photoelectric effect results i n the total attenuation of the photon.  Chapter 1. Introduction  1.4.2  RAYLEIGH  7  SCATTERING  Rayleigh scattering is an interaction between a photon and an atom whereby the photon is scattered by the atom as a whole. T h i s type of elastic scattering results i n essentially no energy being transferred to the atom.  Given that there is no energy imparted to  matter i n this process, Rayleigh scattering is not a concern to Internal Dosimetry but the scattering itself may have an effect on the quality and quantitation of emission data.  1.4.3  COMPTON  SCATTERING  In C o m p t o n scattering the interaction is between the incident photon and a loosely bound electron resulting i n a transfer of energy to the electron. In this case, the incident photon energy greatly exceeds the binding energy of the electron. T h e photon, which has been scattered from its original path, may continue to interact by any of the methods mentioned here. C o m p t o n scattering, as its name suggests, contributes to the scatter of the photons.  The physical characteristics of radionuclides used i n Nuclear Medicine are well known. The emissions that result from the decay of a particular radionuclide are important in determining its useful applications.  Given the short range of the alpha and beta  particles, they become useful tools i n the applications of radionuclides i n therapy - when the destruction of cells is important. T h e high penetrating power of gamma emissions makes them ideal i n diagnostic imaging to provide functional information about a patients specific organ or body system, while simultaneously m i n i m i z i n g the dose to the patient. For therapeutic applications it is also useful to make use of a radionuclide, not only with short range radiation, but also w i t h gamma emissions to aid i n determining dose distributions w i t h i n the body.  Chapter 1.  1.5  Introduction  8  ATTENUATION OF PHOTONS  Quantification in S P E C T requires that the image be corrected for the attenuation of gamma photons that are emitted during the nuclear decay of a radionuclide. These photons interact with tissues in the body. It is during these interactions that the photons may be deflected from their original path and lose some of their energy. Attenuation of photons w i l l differ depending on the energy of the photon and the interacting medium ie. bone, soft tissue, air, etc. A denser m e d i u m (e.g. lead, bone) w i l l attenuate more photons than a less dense medium (e.g. air, soft tissue). Photons with higher energies w i l l have a higher penetrating capability (less attenuation) than photons w i t h lower energies (more attenuation) w i t h i n the same medium.  1.6  GAMMA  CAMERA  Once a radiopharmaceutical has been administered, it is necessary to detect the gamma emissions i n order to obtain the desired functional information. T h e instrument used i n Nuclear Medicine for the detection of gamma rays is known as the G a m m a camera. The G a m m a camera is designed to have good detection efficiency of photons in the energy range relevant to the radionuclides used in Nuclear Medicine (80keV to 300keV) and its efficiency gets poor for energies above this. T h e components of the G a m m a camera are the collimator (limits the acceptance angle of 7 rays onto the detector), the detector crystal (which produces visible photons from the interactions of the 7 rays with the crystal), and an array of photomultiplier tubes on the back surface of the crystal (to transform light into electric pulse and amplify the signal) and position logic circuits (to determine the location of each visible photon as it is produced w i t h i n the crystal).  Chapter 1.  Introduction  9  Organ with radioactive emissions  Zl  Collimator Detector Crystal  Photomultiplier Tube Array Position Logic Circuits  Detector Cover  Figure 1.1: Schematic representation of the Gamma Camera and its components (not to scale). 1.6.1  COLLIMATOR  The detector crystal lacks directional resolution thus a collimator is used to serve this purpose. The collimator is a pattern of holes through gamma ray absorbing material, usually lead or tungsten, which limits the passage of 7 rays to the detector crystal. The collimator achieves this by only allowing those 7 rays traveling within a small range of angles about the normal to the detector face to reach the detector crystal, absorbing those that exceed this angle of acceptance. The absorption of 7 rays in the collimator inherently reduces the sensitivity of the camera by several orders of magnitude. More lead  Chapter 1.  Introduction  10  is used in the collimator for higher energies (to prevent septal penetration of high energy photons) or higher resolution but this reduces the sensitivity of the camera. Less lead can be used at the risk of septal penetration to give the advantage of better photon statistics. The most common collimator i n Nuclear Medicine is the parallel hole collimator owing to its advantageous imaging properties i n that it maintains object/image size. View of collimator from the top:  View of collimator from the side: incoming gamma photons  /  Figure 1.2: Top and Side views of the parallel hole collimator frequently used i n Nuclear Medicine. In the Side view it is apparent that 7 rays having an incidence angle greater than the collimator acceptance angle w i l l be absorbed by the collimator.  Chapter 1.  1.6.2  Introduction  DETECTOR  11  CRYSTAL  W h e n a gamma photon interacts with the detector it does so by means of the Photoelectric Effect or C o m p t o n Scattering w i t h the iodide ions of the crystal [5]. T h i s interaction causes the release of electrons which in turn interact w i t h the crystal lattice to produce light, a process known as scintillation. T h e amount of light produced is directly proportional to the energy lost by the incident photon. A Thallium-activated S o d i u m Iodide [Nal(Tl)] detector crystal is generally used in Nuclear Medicine cameras owing to its good detection efficiency of photons. T h e thickness of the detector crystal is determined by the energy range for which it w i l l be used, generally 3/8-1/2" for the 80-300keV photons in Nuclear Medicine. There is a delicate balance between the detection efficiency of photons and the spatial resolution of the crystal. Thicker crystals manage to detect more gamma rays thus increasing the detection efficiency of the crystal at the cost of decreased spatial resolution. T h i n n e r crystals have poorer sensitivity but compensate w i t h superior resolution. T h e N a l crystal has the capacity for energy discrimination. T h i s allows data collection to be restricted to specific energy windows and also enables the user to collect data i n scatter windows if a scatter correction is desired.  1.6.3  PHOTOMULTIPLIER T U B E A R R A Y A N DPOSITION LOGIC  CIRCUITS  A photomultiplier tube absorbs the light from the crystal on its photocathode and generates about 1 electron for every 7 to 10 scintillating photons absorbed [5]. Following the cathode is a series of dynodes that continually m u l t i p l y the electrons upon colliding w i t h each successive dynode resulting in a signal that is amplified by a factor of ~ 10 . The 6  output signal at the anode is proportional to the number of electrons that were incident at the cathode. Immediately following the photomultiplier tube array are position logic  Chapter 1. Introduction  12  circuits which determine the location of each scintillation event as it occurs i n the crystal.  1.6.4  COMPUTER  The computer stores a l l detection, location, and energy information for each count as it occurs i n real time. T h i s information is stored i n a predefined m a t r i x size ranging from 64 x 64 (pixels) to 256 x 256 for S P E C T projections and 256 x 1024 for a projection scan of the whole body. T h i s projection information can then be displayed using image display software a n d / o r the projections can be input into an image reconstruction software to produce a 3-dimensional representation of the object.  CHAPTER 2 INTERNAL  DOSIMETRY  Once a radiopharmaceutical is administered to the patient it gets distributed throughout the body reflecting, it is hoped, the functioning of a particular organ or to seek out malignant cells for the purpose of therapy. W h e n the affixed radionuclide decays, radiation specific to the particular radionuclide is emitted. The types of radiation of greatest interest in Nuclear Medicine were discussed i n Chapter 1. G i v e n that radiation is being emitted in a biological system, it is important to assess the effect to l i v i n g tissue resulting from this radiation, in order to assess the risks to the patient, or the  effectiveness  of therapy, [6]. T h i s is done by measuring the internal absorbed dose. Specifically, the internal absorbed dose is defined as the amount of energy deposited by ionizing radiation in tissue per unit mass of tissue. Before an explanation can be made of how the absorbed dose is measured, it is wothwhile to describe the theory behind how it is calculated.  T h e Medical Internal  Radiation Dose ( M I R D ) Committee of the Society of Nuclear Medicine set out to provide a means of accomplishing this calculation in 1968 [7], and has been improving upon it ever since.  13  Chapter 2. Internal Dosimetry  2.1  14  T H E MEDICAL INTERNAL RADIATION DOSE ( M I R D )  PROTOCOL  T h e Medical Internal R a d i a t i o n Dose ( M I R D ) Committee developed a simple, yet useful, approach for estimating absorbed doses to normal organs and the whole body from radiopharmaceuticals [8]. T h i s protocol has proven to be a valuable foundation i n internal dosimetry. I w i l l discuss the methods involved i n estimating the internal absorbed dose by the M I R D protocol, the consequential limitations imposed by it, and recent research geared towards compensating for these faults. In order for the absorbed dose to be calculated, two important factors must be determined: the amount of activity administered to the patient, and the length of time the radionuclide resides i n a particular region of the body (in order to calculate the amount of radiation energy deposited at that site and delivered to neighbouring structures over the time that the radiation is present). T h e types of emissions for the particular radionuclide must be identified and, whenever possible, information on the metabolic response for the radiopharmaceutical should be considered. T h e M I R D Committee has attempted to address these issues thoroughly as they pertain to absorbed dose calculations and a description of their methods w i l l ensue. T h e following information is derived from M I R D publications, specifically the MIRD Pamphlet's  2.1.1  Primer  For Absorbed Dose Calculations^},  No.l (Revised)[8\, iVo.5[10], No.lO[Z}, and  PHYSICAL AND BIOLOGICAL  MIRD  No.ll[ll}.  HALF-LIVES  The physical decay rates of radionuclides are constant and well known for commonly used isotopes [12] [13] [3]. W h e n a radionuclide is introduced into the body, to estimate the total time it w i l l remain i n the body, we must also consider the biologic decay of the nuclide. T h i s is because the radionuclide can be eliminated from the body by means of metabolic processes as well as physical decay.  Chapter 2. Internal  Dosimetry  15  The effective decay constant is the total probability that a particular radionuclide w i l l be eliminated, either through radioactive decay or metabolic processes, and is simply the sum of the physical and biological decay constants:  where A denotes the decay constant and the subscripts 'eff', 'p', and 'b' denote effective, physical, and biological, respectively. A half-life can be introduced by: In 2  T  where Ti = half-life, physical or biological. A l t h o u g h the biologic half-life is not as well defined as the physical half-life, it is still possible to model a metabolic rate for a particular radiopharmaceutical [14]. T h e effective half-life of a radiopharmaceutical,  T ff, e  i n a living system can then be determined by  combining the physical half-life w i t h the biological half-life. 1 T ff e  2.1.2  S O U R C E  A  N  D T A R G E T  1 T  p  1 T  b  O R G A N S  Radiopharmaceuticals may deposit themselves i n a number of organs.  T h e organ (or  region) of interest for which the absorbed dose is to be calculated is defined as the target organ. Organs containing radioactivity that contribute to the absorbed dose in the target organ are considered source organs. Specifically, all organs are considered to be source organs if they contain concentrations of radioactivity that exceed the average concentration i n the body. The source and target may be the same organ and, frequently, the most important contributor to radiation dose is radioactivity contained w i t h i n the target organ itself [4].  Chapter 2. Internal Dosimetry  I  16  I - Target Organs  1 - 1 - Source (and Target) Organs Figure 2.1: Concept of source and target regions for internal dosimetry calculations. Source regions (shaded regions) contain activity and may contribute to the absorbed dose i n target regions. 2.1.3  CALCULATING  THE ABSORBED  DOSE  As mentioned previously, the radiation absorbed dose ( D ) is defined as the energy deposited by ionizing radiation ( A E ) i n a mass ( A m ) :  D=  —  Am  The energy released by the radionuclide i n each decay is constant and has been well established.  T h e total energy released i n a target region is the product of the  energy released per decay and the number of decays occurring w i t h i n that target region.  Chapter 2. Internal Dosimetry  17  A l t h o u g h this defines the amount of energy released by the isotope, this may be distinctly different from the energy deposited in the target region. T h i s difference comes from the types of decay emissions, whether they be penetrating or non-penetrating emissions (as discussed in Chapter 1).  The types of emissions determine if they w i l l deposit their  energy within the target region or escape and deposit their energy elsewhere. Hence, the total energy absorbed by the target region is the product of energy incident upon, or released i n , the target region and the fraction that is absorbed [15]. T h e basic absorbed dose calculation, as derived by the M I R D Committee, provides the mean absorbed dose D(r  k  <— 77J to the target region r  k  from the uniformly distributed  activity in the source region r : h  D{r  k  f-  r)  M  h  k  where Ah — the cumulated activity in the source region  (in other words, the total number  of nuclear decays occuring in the source region) M  k  = the mass of the target region r  k  Ai = the radionuclide-specific equilibrium dose constant for radiation type i (or the average energy emitted per nuclear decay i n the form of radiation i) 4>i{ k <— fh) = the absorbed fraction in target region r r  k  source region  for radiation i emitted in  (or the fraction of energy of radiation i emitted i n the source region r  h  that is absorbed i n the target region r ) k  B y the following definitional [fk  h) — the specific absorbed fraction in the target region r  r  emitted in the source region  k  for radiation i  (or the fraction of energy of radiation i emitted i n the  source region that is absorbed per unit mass i n the target region);  Chapter 2. Internal  Dosimetry  18  <f>ii k <r- r ) r  Mk  h  <- r )  r  h  the equation for the mean absorbed dose can be simplified to:  D(r <- r ) = A ^2 &i$i{r <- r ) The M I R D protocol has combined a l l physical factors necessary for dosimetry into k  h  h  a set of factors called the S-factors.  k  h  T h e S-factor has a value for each source/target  combination for the radionuclide of interest. S(r  <- r )  k  region r  h  h)  = the radionuclide-specific S factor for the target region r  and source  k  or the absorbed dose to the target region per unit cumulated activity in the  source region;  S(r  <- r )  k  h  thus D{r  <- r ) = A S{r  k  h  h  <- r )  k  h  B y summing all the absorbed dose contributions from a l l source regions r  h)  the total  mean absorbed dose to the target region r can be estimated: k  D(r ) = YiAS(r +- r )} h k  k  h  This more condensed representation of the total mean absorbed dose D(r ) k  to the target  region is expressed as a sum of the products of two important factors required for this calculation: the cumulated activity A  h  S-factor S(r  k  from a potential source region, and the M I R D  <— rh) which represents the absorbed dose to the target region per unit  cumulated activity in the source region. Most of the biological information (e.g. uptake,  Chapter 2. Internal Dosimetry  19  retention and washout) needed for dosimetry estimations is embodied i n the quantity Ah,, while the physical and anatomical data is included i n the S-factor [16]. Use of the S tables, along w i t h knowledge of the cumulated activity, enables the straightforward calculation of the radiation absorbed dose.  CUMULATED  ACTIVITY  Once the activity A is administered to the patient, it is the effective decay constant X ff 0  e  that characterizes the reduction of activity over time A(t) w i t h i n a source region  A(t)  = ,4  0  exp  The amount of activity i n a source region changes w i t h time and this trend can be portrayed on a T i m e - A c t i v i t y curve where the activity i n the region of interest Ah is plotted as a function of time. T h e cumulated activity, corresponding to a time period from ti to t , is determined by integration from time ti to time t 2  2  of the time-activity  data Ah{t), as determined by serial measurements:  To determine the total absorbed dose, however, it is necessary to integrate from time ti = 0 to time t = oo: 2  The cumulated activity is essentially a measure of the total number of radioactive disintegrations occuring during the time that the radioactivity is present i n the source organ. T h e units of cumulated activity can be expressed i n Bq • s or i n mCi • hr where lmCi is equal to 3.7 x 10 Bq. 7  Chapter 2. Internal Dosimetry  20  M I R D S-FACTORS  S-factors, or the absorbed dose per unit cumulated activity, depend on the mass of the target organ M , the mean energy emitted per disintegration A j , and the fraction of k  energy emitted from the source organ which is absorbed by the target organ 0, [17]. Calculation of the radiation dose for non-penetrating radiations is relatively simple given that /3 particles, a particles, conversion electrons and X and 7 rays of energies less than l l k e V are absorbed in tissue w i t h i n a volume of ~ l c m radius [18]. T h e calculation of radiation dose for penetrating radiations (ie. higher energy photons) can get complicated. Estimates of the S-factors for photons have been calculated by means of Monte Carlo simulations [15] using a geometrical representation of the human body. T h e original 70kg "standard m a n " model, developed by Walter Snyder [19], defined the three-dimensional coordinates of the organs i n this anthropomorphic phantom.  T h e absorbed fractions  for different sources and target organs were then calculated by Monte Carlo analysis of the trajectories of large numbers of photons of different energies [15] [20]. Since the original "standard man" model, the M I R D protocol has also developed S-factors based on pediatric phantoms for newborns and ages 1 year, 5 years, 10 years, and 15 years-old, the adult female phantom (based on the 15 year-old male phantom) [21], and the pregnant female at 3, 6, and 9 months gestation [22]. It is crucial to keep i n m i n d that the shapes and proportions of the phantoms used to calculate S-factors only crudely approximate human anatomy. A c t u a l patient values may vary considerably from model parameters, especially i n the case of patients w i t h various levels of disease.  CHAPTER 3 MEASUREMENT TECHNIQUES  In this chapter I w i l l address the practical question of How is the absorbed dose measured? The S-factors have already been tabulated for all diagnostic and therapeutic radionuclides used in Nuclear Medicine. T h e majority of effort i n dosimetry studies is spent determining the cumulated activity A  h  for each potential source organ. A l t h o u g h there is a common  basis for a l l methods used to find the cumulated activities, a variety of approaches have been proposed in hopes of achieving more accurate results. Before delving into the most common method for determining the cumulated activity, it is necessary to understand the imaging techniques available for data collection specifically planar and S P E C T imaging - and the effects of attenuation in imaging. Once this foundation is established, I will summarize the current methods used i n dosimetry, how quantitation is achieved i n planar and S P E C T imaging and, finally, the dosimetry method proposed i n this work.  21  Chapter 3. Measurement Techniques  3.1  3.1.1  IMAGING  PLANAR  22  METHODS  IMAGING  Planar imaging (represented i n Figure 3.1 (a)) uses a stationary g a m m a camera, positioned close to the patient, to detect gamma emissions that originate w i t h i n the patient. T h i s type of imaging yields a two-dimensional representation of a three-dimensional activity distribution. T h e collimator, as described in Chapter 1, allows only photons travelling within a particular range of trajectories (the acceptance angle) to interact with the detector crystal, absorbing those photons that exceed the acceptance angle. The photon's direction of travel and its point of interaction within the detector crystal define a line somewhere along which the emission took place, therefore several layers of the body have their contributions superimposed. Planar imaging is limited by its lack of depth localization of the emission origins.  3.1.2  SPECT  IMAGING  Single P h o t o n Emission C o m p u t e d Tomography ( S P E C T ) , schematically represented i n Figure 3.1 (b), acquires a set of 2-dimensional planar images (projections) by rotating the camera around the patient and acquiring data at a number of different angles (views). The multiple views around the patient allow for the reconstruction of a 3-dimensional representation of the activity distribution i n the body.  3.1.3  A T T E N U A T I O N E F F E C T S IN IMAGING  W h e n the gamma ray photons travel through the patient, there is a p r o b a b i l i t y that the photons w i l l interact with the tissue, either by scattering or absorption, resulting in photon attenuation.  T h e deeper in the patient the source of gamma rays is, the more  those gamma rays are attenuated on the way to the detector. A t t e n u a t i o n is nonuniform  Chapter 3. Measurement Techniques  23  Figure 3.1: Schematic representation of (a)planar and ( b ) S P E C T (2 detector heads) imaging and the projection profiles collected by the detectors. P l a n a r and S P E C T projections are 2-dimensional representations of a 3-dimensional object. T h e ability of S P E C T to rotate around the object collecting multiple 2-D views enables the reconstruction of a 3-D image of the object. throughout the body since different tissues exhibit different aptitudes for attenuating photons (e.g. high attenuation i n bone and low attenuation i n the lungs). Since attenuation w i l l result in an image that misrepresents the real distribution of radiation within the body, data should be corrected for attenuation otherwise some areas of the patient w i l l appear to be less radioactive than they really are.  LINEAR ATTENUATION  COEFFICIENT  W h e n a number of photons N passes through a material of thickness Ax, n number of photons will interact with the attenuator and be removed from the initial flux of photons, this removal of photons can be represented by  n — (j,NAx  Chapter 3. Measurement Techniques  24  The factor /J is the Linear Attenuation Coefficient and characterizes the attenuating properties of the material it represents.  Specifically, u. is a constant that describes the  number of photons attenuated per unit length of material for which the photons traverse. Not only does p depend on the attenuating material it describes, but i t also depends on the energy of the photons. T h e relationship between the number of incident photons N  0  on a material having a linear attenuation coefficient p for that photon energy, traversing a thickness x of the material and resulting i n N photons successfully transmitted through that thickness of material without being attenuated, is:  N = N exp-^  x  0  T h i s equation describes the attenuation of photons by any thickness of material.  3.2  CURRENT METHODS  FOR DOSIMETRY  The M I R D protocol has the limitation of basing its calculations on 'standard' anatomical geometries - ignoring the importance of individual patient differences along w i t h anatomical mutations associated with some diseased states. Suggestions have been made, and protocols developed, i n an effort to improve upon the M I R D protocol. Most of these techniques are supplements for improving the M I R D protocol rather than complete new protocols. It has to be recognized that the time and effort put into absorbed dose calculations by the M I R D committee would be difficult to surpass. It is therefore wise to take advantage of the insights the M I R D protocol can provide, limited as it may be. W i t h the introduction of new radiopharmaceuticals, specifically monoclonal antibodies ( M A B s ) designed to target antigens generally associated w i t h malignant cells, the activity is no longer localized i n specific volumes. T h i s activity spread, along with localized uptake i n tumors, hinders the straightforward application of the M I R D protocol.  Chapter 3. Measurement Techniques  25  This is due to the fact that the S-factors have be designed for standard organ systems as sources/targets of radiation. Whether or not a tumor resides i n a normal source region, it should be treated as a separate volume i n dosimetry calculations. T h e most basic technique for tumor dosimetry is obtained by considering only the electron self-dose, or if the tumor is assumed to be a sphere or ellipsoid then the self-dose photon contributions can also be included by using the absorbed fractions listed i n M I R D Pamphlet No.5 [15]. These techniques do not provide dose contributions to normal tissue from activity i n the tumor nor do they include the tumor dose from activity i n normal tissues. A software program MABDOS  designed by T i m o t h y Johnson [23] has overcome this predicament by  recognizing non-standard volumes positioned i n a non-standard geometry. M A B D O S accomplishes this by performing on-the-fly Monte Carlo simulations to determine S-factors for tumors as source and target organs, whereby the tumor is treated as a spherical perturbation to the Standard M a n geometry [23]. Patient-specific approaches to dosimetry generally require a 3-dimensional representation of patient anatomy ( C T or M R I ) as well as the radioactivity distribution within the patient ( S P E C T , P E T , or biopsy samples). These patient-specific methods can be categorized as either Monte Carlo [24] [25] or point-kernel [26] [27] [28] [29] based. Monte Carlo methods provide simulation of particle transport and a tallying of energy deposited in target regions. T h e Monte Carlo code can use a S P E C T or P E T activity array to determine the number of radionuclide emissions at each array unit [26], while the emissions can be followed across a C T density array [25] to determine points of absorption in the patient or escape. Point-kernel consists of tables of absorbed doses versus the corresponding distances from a point source. It incorporates the emission spectrum of the radionuclide and its distribution ( S P E C T or P E T ) and the absorbing medium ( C T or M R I ) . T h e distance between a given voxel inside an activity-containing source volume and a point  Chapter 3. Measurement Techniques  26  on the target plane is calculated, while always considering the traversing medium. T h e distance is then found i n the lookup table of dose versus distance (point-kernel) to obtain the dose per unit cumulated activity [27]. The majority of these dosimetry revisions have focused on ways to improve absorbed fractions i n the presence of tumors. Whether they were developed to be patient-specific or not, they have a l l focused on ways to improve the S-factors, or some component of the S-factors. Seldom is there a suggestion to improve upon methods used to measure the cumulated activity. Determining the cumulated activity v i a excreta counting (e.g. urine and feces) is not sufficient as the sole method of data collection and should only be used as a supplement to other methods. Tissue sampling (blood or biopsy) is usually not a viable option due to its invasive nature and may lack accuracy i f used as a complete method. External non-imaging radiation monitoring (e.g. radiation counter) [30] is useful for determination of whole-body activity but lacks the capacity to deliver dose estimates for specific regions. Quantitative planar imaging makes possible the determination of organ and tumor activities but, due to its 2-D nature, can yield high errors i n dose estimates not suitable for therapy. The most notable recommendation to revise cumulated activity measurements has been the proposal of quantitative S P E C T scans rather than planar scans [31]. T h i s may be an attractive alternative i n light of quantitative improvements i n S P E C T but it may not be realistic considering the time required for one S P E C T scan, let alone the multiple scans necessary i n dosimetry studies. It has also been proposed to implement only one S P E C T scan and use this as a constraint for the planar scans [32]. A slight modification to this proposal is the inspiration for this thesis. Because this thesis attempts to improve upon the determination of the cumulated activity A^, it may be complemented by the above attempts to include S-factors by non-standard regions. Together these methods  Chapter 3. Measurement Techniques  27  could have the potential to generate more accurate estimates of the absorbed dose.  3.3  QUANTITATIVE  3.3.1  ATTENUATION  PLANAR AND S P E C T  CORRECTION  IN P L A N A R  IMAGING  IMAGING  The conventional manner i n which attenuation correction is carried out i n planar imaging is invariably crude yet swift. T h e general protocol i n internal dosimetry involves whole body planar scans of the patient from both anterior and posterior views. Since the data in both the anterior IA and posterior I projections are less than the unattenuated data P  they do not accurately represent the unattenuated activity distributions. However, a mean image may be modified by a correction factor to compensate for the attenuation. B y assuming a constant attenuation coefficient i n a region, a hyperbolic sine correction can be used [33]. Figure 3 . 2 represents the source distribution w i t h i n an attenuating region to supplement the proof for the hyperbolic sine correction. The anterior 1^ and posterior Ip projections can be represented by:  Lk  where K — the linear concentration of the isotope activity i n the source region, assumed constant / = the linear fraction of the thickness i n which the isotope is distributed m = the mean depth of the source distribution L = the total thickness through which the ray passes  Chapter 3. Measurement Techniques  28  ANTERIOR PROJECTION  L  0  POSTERIOR PROJECTION  Figure 3.2: T h e anterior and posterior projections from a constant source distribution K (shaded object) at a mean depth m within the constant attenuator p. p = the linear attenuation coefficient for the attenuating object Integrating a projection (posterior, for example) over the source region:  Jm-£±  From the definition sinhA = \{e  A  p  L  ™-4r  J  — e~ ) the integrated posterior projection can be A  simplified to: 2Ke~»  m  IP = Likewise for the anterior projection:  p  . sinni  pfL 2  )  Chapter 3. Measurement Techniques  I  A  29  =  sinh{——) £t  jJL  The geometric mean of these two projections yields: r— 2Ke'f \II Ip = v  A  . fxfL sinh{——) p i  Since the unattenuated projection data P is simply P = KfL,  rearranging and substi-  tuting for K gives an expression for the corrected projection: /xfLe ^ 1  =  2  y/I Ip A  sinh(^)  It is important to note that this correction is independent of source depth ra. A similar process for an arithmetic mean yields the following: _^fLef  (IA + IP)  r  4  sinh(^)cosh(^  - ra))  The corrected projection using an arithmetic mean is dependent on source depth. T h e extra factors required to complete the calculation, generally accompanied by a lack of knowledge of the source depth, makes using an arithmetic mean misguided and undesirable. To apply theory to practice, the expression for the corrected projection is simplified by combining —rhrrcr into a factor  F; is a correction for the source region attenuation  coefficient (pj) and source thickness ( / L ) [34] and w i l l not deviate significantly from 1.0 until ix j or fL becomes large (unlikely for most source regions unless a significant background of activity is present).  A calibration source of known activity C must be  placed i n the field of view for each scan i n order to convert image counts to activity. In a typical dosimetry (imaging) study, the first set of scans is taken immediately postinjection and the following scans are taken over time until the activity i n the body has  Chapter 3. Measurement Techniques  30  dropped to no more than 10% of the initial activity. F r o m these scans a region of interest (ROI) is drawn around any region that contains a concentration of activity (counts) that exceeds the average concentration i n the body. T h e anterior, 1^, and posterior, I , P  counts are then averaged to attain a mean count for that R O I . T h i s method then takes the patient's anterior/posterior thickness L (at the waist or chest) and assumes this as the attenuating thickness. to that of water.  T h e attenuating medium is generally assumed comparable  Using half of the anterior/posterior thickness, | , and an effective  attenuation coefficient equal to that of water, p, for the 7 ray energy of interest (and energy window), a correction is then made to increase the counts by a factor of  .  The counts are then converted to activity by means of the calibration source R O I i n the image. For a geometric mean, the planar attenuation correction can be represented by the following equation [34]:  where Aj = the attenuation corrected activity i n the source region C = calibration factor (count rate per unit activity) The planar attenuation correction using an arithmetic mean is represented by: Fj{I  + I ) ±L  C  2  —- 1 J  A  P  Lp  2  These corrections consistently overestimate the counts i n the thinner regions of the patient and ignore the important contributions of differing tissue densities w i t h i n the patient.  T h i s over-correction obviously has the safety of the patient i n m i n d but, as  we will see later, such an overestimation could prevent effective therapy from being administered.  Chapter 3. Measurement Techniques  31  W h e n all the activities are determined for a particular R O I , they can then be plotted as a function of time (post-injection) on a T i m e - A c t i v i t y curve. Figure 3.3 is an example of a T i m e - A c t i v i t y curve. Sample Time-Activity Curve  100  200  300 Time (minutes)  400  500  600  Figure 3.3: Sample T i m e - A c t i v i t y curve as used to determine the cumulated activity. A function can then be mathematically fit to the data points, generally described by a mono-exponential (general clearance of the radiopharmaceutical), bi-exponential (slow and fast clearance rates), uptake-clearance bi-exponential (uptake and clearance rates), among others.  T h e fit depends on the behavior of the radiopharmaceutical w i t h i n the  patient and the source region. Once this function is attained it is possible to extrapolate the behavior of the radiopharmaceutical to time t = oo. If the scans are taken short of  Chapter 3. Measurement Techniques  32  the radiopharmaceutical decaying to 10% of its original activity, then more error may be introduced when calculating a fit to the data. Unless the behavior of the radiopharmaceutical has been well established then it is common protocol to continue scanning the patient until this 10% mark has been achieved. The area under the time-activity curve (function fit to the data points) yields the cumulated activity A h necessary to complete the absorbed dose calculation. Recalling from the previous chapter, this area is determined by integration of the function A h from time t — 0 to t = oo:  3.3.2  ATTENUATION  CORRECTION IN  SPECT  From no attenuation correction at a l l to the crude method of approximating the patient as an ellipse of uniform attenuation [35] to the refined methods of today - attenuation correction i n S P E C T has evolved greatly. Patient-specific methods of attenuation correction generally make use of a point, line, or flood transmission source to determine the patient's internal density distribution. T h e method of attenuation correction that will be discussed i n this thesis makes use of a M u l t i p l e Line A r r a y ( M L A ) transmission source developed by the M e d i c a l Imaging Research G r o u p ( M I R G ) [36] at Vancouver Hospital & Health Sciences Centre i n collaboration w i t h Siemens.  T h i s transmission source is  made up of a series of parallel collimated Gadolinium-153 line sources. T h e sources are placed i n a manner whereby the strongest sources are placed i n the middle and each successive source has reduced activity. T h e ratio between each neighbouring pair is constant throughout the array (from the middle outwards). Figure 3.4 depicts the concept behind the M L A transmission source. The M L A is mounted opposite the detector head and undergoes the same rotation,  Chapter 3. Measurement Techniques  a)  TRANSMISSION SOURCE  33  b)  TRANSMISSION SOURCE  Figure 3.4: M L A transmission source and resulting profiles (a)without and (b)with the attenuating object. always keeping its face to the detector. T h e Gd-153 line sources i n the M L A decay by means of electron capture, resulting i n the emission of 7 rays. T h e fundamental emissions from the Gd-153 source are 9 7 K e V and 103keV 7 rays producing a peak emission profile centered about lOOkeV. A t each projection, emission data from w i t h i n the patient are acquired into one energy window (e.g. centered around 140keV for Technetium-99m) simultaneously w i t h transmission data collected into the lOOkeV transmission window. Simulated emission/transmission data is represented i n Figure 3.5. A scan is taken of the Gd-153 transmission source without attenuating objects i n the field of view i n order to compare these to patient-inclusive projections. T h e ratio between the transmission projection without (3.4 (a)) and with (3.4 (b)) the attenuating object is used to reveal the effects of the attenuation for each projection. T h e transmission data for a l l projections can then be reconstructed to yield a 3-dimensional attenuation map of the patient.  Chapter 3. Measurement  Techniques  34  Transmission Energy  E m i s s i o n Energy Window  Window  Counts  100  140  Energy (keV) Figure 3.5: Energy spectrum of the Gd-153 (transmission) and Tc-99m (emission) sources used in a simultaneous emission/transmission scan. 3.4  D O S I M E T R Y M E T H O D P R O P O S E D IN THIS  WORK  This thesis proposes a protocol that builds on the quantitative planar protocol commonly used today. Like the planar protocol, multiple anterior and posterior planar scans (with the presence of a calibration source) are taken at regular intervals until the activity has sufficiently decayed (ie. where the errors introduced by extrapolation of the decay function can be minimized). Once the counts for the R O I in the anterior view IA and posterior view Ip are averaged by a geometric mean, and converted from counts to activity by means of the calibration source C, the e 2 attenuation correction is applied. The attenuation corrected activities Aj for the source region j are plotted on a timeactivity curve and an appropriate function is fit to the line. In addition to the quantitative planar protocol, a m i n i m u m of two quantitative S P E C T scans are acquired with the source region of interest in the field of view of  Chapter 3. Measurement Techniques  35  the camera. B y quantitative S P E C T I imply a simultaneous emission/transmission scan using the Siemens M L A transmission source [36]. T h e presence of a calibration source is also required i n order to convert counts to activity. Siemens Profile software uses the patient-specific attenuation map to iteratively reconstruct a 3-dimensional attenuation corrected image of the activity distribution w i t h i n the patient. T h i s 3-D image can then be imported into Display software, produced by Christine D y k s t r a [37], whereby contiguous volume analysis is performed on the data to seek out 3-D regions of interest. T h e counts i n the 3-D source R O I ' s can then be compared to the counts i n the 3-D calibration R O I to make the conversion from counts to activity. More than one S P E C T scan is preferential i n the protocol, not only to ensure that no gross errors have occurred during one of the scans, but to maximize the benefits of S P E C T quantitation. T h e 3-D source region activity, determined from the S P E C T image, is accepted as the true activity i n the region, hence it is used to constrain the planar data points. T h e term constrain refers to a normalization factor applied to the planar data. A normalization factor N, specific to each region, is determined by the ratio of what the planar function A(t) would yield as an activity for the time the S P E C T image was acquired (A(t = SPECT))  to the true activity i n the 3-D source region A(t ) true  as  determined by the S P E C T image. _ A{t =  SPECT) Atrue  A normalization factor can be calculated for every S P E C T scan. A mean normalization factor can then be used to constrain the planar data. A l l planar data points are then divided by the normalization factor N (ie. constained), to yield the S P E C T - c o r r e c t e d data. A function is fit to the new data points so that the cumulated activity A can be detemined by integrating the function from t = 0 to t = oo. Once the cumulated activities are calculated for all potential source organs, the absorbed  Chapter 3. Measurement  Techniques  36  dose can be calculated by using the M I R D S-factors, or some other proposed method for determining S-factors as suggested earlier i n this chapter.  CHAPTER 4  SIMULATIONS  Mathematical simulations were performed i n order to evaluate the errors introduced by different methods of image averaging and attenuation correction i n planar imaging techniques. The first technique, performed using Matlab, was simplistic i n nature, lacking a realistic portrayal of photons i n matter, yet not completely devoid of useful information. Even though the trajectories of the photons are not representative of their true behavior (ie.  interactions on the path to the detector are ignored), the calculations performed  on the collected photons are realistic in the systematic errors they reveal. T h e second technique, performed using SimSET or Simulation System for Emission Tomography (a Monte Carlo photon history generator), realistically portrays the interactions of photons through the attenuating material, any absorptions or energy losses of the photons, the collimation and detection system, and the energy window employed to detect the photons. A g a i n , these simulations are used to assess the errors incurred by planar attenuation correction.  37  Chapter 4.  4.1  Simulations  MATLAB  38  SIMULATIONS  To describe the attenuating object, the M a t l a b simulations utilized a single slice of an elliptical object (diameters of 20cm by 40cm) to model a cross-sectional view through a patient's mid-section.  oI  1  1  1  1  1  1  10 -  20 r  30 h  40 h  50 -  60 1  0  1  10  1  20  30 nz = 632  1  1  40  50  1  60  Figure 4.1: Pixelized representation of the attenuating ellipse used i n M a t l a b simulations.  Different combinations of one, two, or three spherical activity sources (each of 4cm diameter) were placed w i t h i n this slice for the case of no background activity as well as for the case of a uniformly distributed background of activity w i t h i n the ellipse. The 1, 2, 3, 4, and 5 notations on the activity sources i n Figure 4.2 denote the activity coordinates of (0,0),  (0,6),  (-16,0),  (8,0),  and  (-8,3)  respectively where the origin  (0,0)  Chapter 4. Simulations  39  10  20  G  30  e  tO  40  50  60 10  20  30  40  50  60  Figure 4.2: Locations of the activity sources and background activity w i t h i n the M a t l a b object. is defined as the centre of the attenuating ellipse (coordinates are expressed in cm). Outside the attenuating ellipse there was assumed to be a non-attenuating medium, approximating that of air. Three separate matrices (each 64 x 64) were created, one to describe the activity indexes per pixel, and the other two to describe the anterior and posterior attenuation indexes for each pixel. A representation of the attenuating matrices is shown in Figure 4.3. E a c h pixel in the activity spheres was specified to have 10000 starting photons and if a background was used then the background pixels were specified as having 100 starting photons per pixel. T h e attenuating m a t r i x defined a weighting for each pixel (based on p = 0 . 1 2 c m  - 1  assuming water equivalency for a photon energy  window of 126-154keV to simulate the most frequently used radionuclide i n Nuclear Medicine, Technetium-99m) depending on the pixel's depth w i t h i n the ellipse relative  Chapter 4. Simulations  40  to where the detector head would be positioned. T h e cumulative pixel weightings along the path from the photon's origin to the detector face determines whether the radiation arising in the pixel w i l l be absorbed w i t h i n the attenuating ellipse or penetrate the ellipse to be counted by the detector. A g a i n , the pixels outside the ellipse d i d not contribute any  attenuation.  ANTERIOR DETECTOR (MATRIX)  /  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 syn 12 12 «. ° ° 0 0 24 24 24 / \ 36 36 36 \ 48 ?,4 60  /  <1  \\  p?4 48 W  \  \  / \  j I  /  \  /  \  r24 48 60 ,48 36 ,36 36 \ V 24 24 24 / 0 0 •ft. 42 ,12 12 » D 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0  1  \  ll2 36 24  \  •  / 0 0 0 0 0  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0  /  POSTERIOR DETECTOR (MATRIX)  Figure 4.3: Anterior and posterior attenuating matrices. A c t u a l matrices are 64 x 64. Each pixel has an attenuation weighting defined by its depth w i t h i n the attenuating ellipse as measured from the detector head.  The  simulations exclude any possible scatter of the photons, assuming that if they are  not completely absorbed on their path to the detector (depending on the pixel weightings along the way) they are collected as a photon count on the detector b i n that is in a direct line from the point of the photon's origin. Photons originating w i t h i n a given pixel are defined as originating in the centre of that pixel. B y definition this means that the photon is only attenuated by half of the originating pixel (ie. half of the pixel weighting).  Chapter 4. Simulations  41  Depending on the activity arrangement defined for each simulation, the activity matrix would then be multiplied pixel-by-pixel by the attenuating m a t r i x and the resulting counts placed i n 64 detector bins (for the single slice simulated). T h e attenuating m a t r i x would obviously be equal but opposite for the anterior and posterior projections. Once the binning was done, the anterior I and posterior I projections were then averaged, A  P  either by an arithmetic or geometric mean, and multiplied by a factor for attenuation correction. T h i s procedure is outlined by the following equation:  = VX^pexpV  C orr c  for a geometric mean and _  Ccorr —  (I  A  +  HL  Ip)  2  exp  2  for an arithmetic mean. Where C  corr  represents the attenuation corrected counts i n the projection, // is assumed  to be the linear attenuation coefficient of water over a photon energy window of 126154keV, centered about 140keV (simulating Tc-99m), and L denotes the attenuating ellipse's anterior/posterior thickness. The linear attenuation coefficient ji for Tc-99m 140keV 7 rays has a table value of 0 . 1 5 c m . T h i s tabulated value fails to account for the fact that i n Nuclear Medicine -1  studies photons are accepted over a range of energies (ie. 126-154keV rather than only 140keV). A tabulated value of \x = 0 . 1 5 c m  - 1  for 140keV i n water indicates that 15%  of the monoenergetic photons w i l l be attenuated i n 1cm of water. Obviously the value of 0 . 1 5 c m  - 1  fails to account for the possibility that some of the attenuated  (scattered)  photons may still be collected by the detector, therefore the tabulated value must be appropriately adjusted.  T h e value is often adjusted to a lesser value generally near  Chapter 4, Simulations  42  0 . 1 2 c m . M a t l a b simulations were carried out for the two cases where p = 0 . 1 5 c m -1  and p = 0 . 1 2 c m  - 1  - 1  to demonstrate the typical errors that can arise.  To calculate the percent error resulting from this method of attenuation correction, the following equation was used: 07 7-1  C  Ct  corr  %Error —  rue  —  x 100  (-'true  where C e represents the true activity originating from w i t h i n the object. trU  4.1.1  RESULTS  Results of the M a t l a b simulations for p — 0 . 1 5 c m  - 1  and p = 0 . 1 2 c m  - 1  are summarized i n  Table 4.1 and Table 4.2 respectively. T h e most conspicuous difference between Tables 4.1 and 4.2 is the magnitude of the errors. Table 4.1 clearly shows larger errors for a l l cases in comparison to Table 4.2. T h i s is simply due to the larger linear attenuation coefficient of p = 0 . 1 5 c m  - 1  rather than p = 0 . 1 2 c m . In a l l cases, the presence of background -1  radiation introduces more error when correcting for attenuation (with the exception of activity source 3 ) . A n y peripheral (off-centre) activity w i l l always be over-corrected due to the fact that the attenuating thickness used to correct for attenuation is selected as half of the thickest w i d t h of the patient (in this case, half of the thickest part of the ellipse, | f = 10cm). T h i s correctly estimates the thickness at the centre of the patient (ellipse) but over-estimates the thickness at the periphery. These results indicate that a tabulated value of the linear attenuation coefficient p clearly produces an overestimation of the true counts i n the object. A n overestimation of this magnitude would translate into a gross overestimate of the cumulated activity (hence absorbed dose) i n the region, potentially preventing adequate therapy from being administered.  Chapter 4. Simulations  43  Table 4.1: M a t l a b results of attenuation correction using planar techniques with / i = 0 . 1 5 c m ( A c t i v i t y Source Combinations are labeled i n Figure 4.2). -1  Activity Source Combination NONE 1 2 3 4 5 1+2 1+3 1+5 2+3 2+5 3+4  Arithmetic Percent Error (%) No Background W / B a c k g r o u n d 94.1 56.0 80.1 111.2 67.3 73.9 62.6 81.4 58.9 95.9 73.4 88.2 92.2  4+5 1+3+4  -• 35.9 72.6 120.1 53.2 63.2 54.3 78.0 49.6 96.4 67.9 86.7 91.7 58.2 -  1+2+3  -  1+3+5 2+3+4  -  78.9 76.2  2+3+5 2+4+5 3+4+5  -  3+5  -  -  65.7 73.4  83.8 86.6 67.7 81.1  Geometric Percent E r r o r (%) No Background W / B a c k g r o u n d 35.9 35.9 120.1 53.2 53.2 44.8 78.0 44.5 78.0 44.5 86.7 86.7 53.2 -  94.1 56.0 57.8 111.2 67.3 67.8 55.4 81.4 55.1 82.5 56.2 88.2 88.5 62.0 73.4 73.7  -  73.6 74.2 74.4 55.4  -  78.5  -  Chapter 4. Simulations  44  Table 4.2: M a t l a b results of attenuation correction using planar techniques w i t h /J. = 0 . 1 2 c m ( A c t i v i t y Source Combinations are labeled i n Figure 4.2). -1  Activity Source Combination NONE 1 2 3 4 5 1+2 1+3 1+5 2+3 2+5 3+4 3+5 4+5 1+3+4  Arithmetic Percent Error (%) No Background W / B a c k g r o u n d  0.7 27.9 63.1 13.5 20.9 14.3 31.9 10.8 45.5 24.4 38.3 42.0 17.2 -  1+2+3 1+3+5 2+3+4  -  2+3+5 2+4+5 3+4+5  -  -  -  43.8 15.6 33.4 56.4 24.0 28.8 20.4 34.4 17.7 45.1 28.5 39.4 42.4 22.8 28.4  Geometric Percent E r r o r (%) No Background W / B a c k g r o u n d -  0.7 0.7 63.1 13.5 13.5 7.3 31.9 7.1 31.9 7.1 38.3 38.3 13.5 -  32.5 30.5 36.2  -  38.3 24.2 34.2  -  -  -  -  43.8 15.6 16.9 56.4 24.0 24.3 15.1 34.4 14.9 35.2 15.7 39.4 39.6 20.0 28.4 28.7 28.6 29.0 29.2 15.1 32.2  Chapter 4. Simulations  45  For a l l cases i n Tables 4.1 and 4.2, the errors produced by the use of an arithmetic mean were either equal to or greater than the errors produced by a geometric mean. It is plausible to explain this difference by the proof i n Chapter 3, whereby attenuation correction by a geometric mean is independent of source depth and the correction by an arithmetic mean is dependent of source depth. These preliminary results tend to favor the use of a geometric mean. A further inquiry into the question of an arithmetic vs geometric mean is presented i n the following set of simulations using SimSET.  4.2  SIMSET  SIMULATIONS  This Simulation System for Emission Tomography uses Monte Carlo techniques to accurately model photon interactions w i t h the attenuating object. T h e attenuating object was defined to be a circular cylinder of 20cm radius (x- and y-axis planes) w i t h a length of 24cm (along the z-axis). T h e length of the cylinder was comprised of 12 slices (2cm each). Each simulation was set to track 1 0 photons (designated as Tc-99m 7 rays of 7  140keV) through their interactions, escapes, and/or detection. T o model a planar scan, detection was performed by simulating detectors at 0° (anterior) and 180° (posterior), with the data for each projection binned into a 128 x 128 matrix. Like the Matlab simulations, an energy window from 126-154keV was used to allow for some scattered radiation - as this 20% energy window is often used i n Nuclear Medicine.  These 10 photons 7  were distributed corresponding to the defined activity geometry for each simulation, and each photon was emitted randomly i n a 47r solid angle. E a c h simulation for a particular activity distribution was performed twice - once with the attenuating cylinder i n the simulation and once without the attenuating cylinder (in order to estimate the true activity to compare with the attenuated case). T h e purpose of these simulations is to compare the results of attenuation corrected data to the unattenuated true data.  Chapter 4. Simulations  46  ANTERIOR PROJECTION  ROI "I  ROI # 2  POSTERIOR PROJECTION  Figure 4.4: Representation of the activity distribution (as viewed i n the central slice) and a l l 12 slices of the attenuating cylinder. Once the anterior and posterior projections were simulated, R O I ' s were drawn around activity regions and the counts from these R O I ' s were noted. T h e count total i n a particular R O I was averaged w i t h its opposing projection counterpart ( R O I count total), either by an arithmetic or geometric mean, then an attenuation correction factor was applied to obtain the corrected counts for that R O I . Rather than using a tabulated p, an effective p was used that incorporated the effects of the energy window and also the attenuating medium (water). T h i s effective p, evaluated as 0 . 1 2 7 c m , was determined by comparing -1  the mean counts i n the attenuated run to the mean counts i n the unattenuated run for the case where there was one activity sphere directly i n the centre of the attenuating cylinder. Simulations were carried out using either one, two, or three activity spheres i n a uniform background of activity or no background activity. T h e positions of the spheres  Chapter 4. Simulations  47  in the central slice of the object are illustrated i n Figure 4.5. T h i s representation lacks any depth element i n the z-direction, although it is implied. E a c h activity source is 4cm in diameter, spanning over the two central slices. If a background of activity is included in the simulation then this activity spans uniformly over the entire cylinder (all 12 slices). Each activity source was given an activity of 0.255mCz per voxel (a voxel being 1 c m ) 3  and i f a background of radiation was included i n the simulation then a l l regions i n the cylinder (excluding sphere regions) were given an activity of O.OOlmCi  per voxel. To  clarify the units, a mCi is equal to 3.7 x 10 Bq (bequerels) where a bequerel is one 7  disintegration of radioactive material per second.  40  60  120  20  40  60  80  100  120  Figure 4.5: A c t i v i t y positions as viewed i n the central slice of the attenuating cylinder used in S i m S E T simulations. The 1, 2, and 3 notations on the activity sources i n Figure 4.5 denote the activity coordinates of (0,0,0), (0,10,0), and (10,0,0) respectively where the origin (0,0,0) is defined to be the centre of the attenuating cylinder (coordinates are i n cm). In the case of a single activity source there is only one R O I to consider i n the projections.  For  Chapter 4. Simulations  48  multiple activity sources (2 or 3) there may be one or two R O I ' s (as there is overlap of the 1 and 3 activity positions). T h e R O I # l refers to the central region of interest and the R 0 I # 2 refers to the region offset i n the y-direction. These results are summarized in Table 4.3. Table 4.3: S i m S E T results of attenuation correction using planar techniques. Activity Source Combination  Arithmetic Percent Error (%) No B K G W / B K G  1 2  0 34  3 1+2  125 5 26 66 60 38  1+3 1+2+3  26 58 135 23 45 65 61 51  Geometric Percent E r r o r (%) No B K G W / B K G 0 34 -1 5 26 30 26 38  25 58 48 23 45 43 35 51  ROI# 1 1 1 1 2 1 1 2  To demonstrate the results of differing levels of background, another set of simulations was carried out using a l l three activity spheres whereby the activity per voxel i n the activity spheres was reduced from 0.255mC? to O.lbOmCi to 0.075mCi, hence increasing the background contribution (of the total activity) from 54% to 67% to 80% respectively. A g a i n , an effective p of 0 . 1 2 7 c m  -1  was used i n the attenuation correction calculations.  These results are summarized i n Table 4.4.  4.2.1  RESULTS  Like the previous Matlab simulations, the results presented i n Table 4.3 demonstrate that the errors produced by the use of an arithmetic mean are either equal to, or greater than, the errors produced by use of a geometric mean.  For this reason, i n the next  Chapter 4. Simulations  49  Table 4.4: S i m S E T investigation of the effects of different levels of background radiation on attenuation correction i n planar techniques. A c t i v i t y Per Voxel In the A c t i v i t y Spheres (mCi)  Arithmetic Percent Error (%) No B K G W / B K G  0.255  67 38  76 57  0.150  58 35 65 27  81 70 89 108  0.075  Geometric Percent E r r o r (%) No B K G W / B K G 34 38 24 35 30 27  44 57 59 69 72 107  ROI# 1 2 1 2 1 2  Chapter on Phantom Experiments, a geometric mean is employed when performing planar attenuation correction.  T h e presence of background radiation also contributes to an  increased error i n the estimate of the activity i n the source region. A s Table 4.4 illustrates, increasing the background level from 54% to 80% has increased the error i n the activity estimate from 44% to 72% i n R O I # l and from 57% to 107% i n R O I # 2 (values are quoted for a geometric mean).  CHAPTER 5 PHANTOM  EXPERIMENTS  P h a n t o m experiments were performed w i t h the intention of obtaining time-activity curves for regions of interest so that the cumulated activity could be calculated and compared to the known cumulated activity. T h e known activity i n the phantom was determined by the accurate calibration of the activity-containing syringes both before and after the contents were injected into specific regions i n the phantom. N o biological decay exists in the phantom hence only the known physical decay constant X of the radionuclide is p  needed to calculate the cumulated activity A i n each R O I . The phantom employed i n these experiments was a thorax phantom, simulating the chest of a patient. T h i s phantom has dimensions of 32.0cm by 23.5cm (diameters), and contains two low-attenuating inserts (to simulate lungs), a high-attenuating insert 37mm in diameter (to simulate the spinal cord), and a water/activity tillable insert (to simulate the heart).  T h e body of the phantom could also be filled with activity to replicate a  background of activity within the body.  T h e second experiment also made use of 2  activity filled tumor inserts w i t h i n the phantom.  50  Chapter 5. Phantom Experiments  5.1  51  TECHNETIUM-99M PHANTOM EXPERIMENT #  1  The radionuclide used i n this experiment was Technetium-99m since it is relatively cheap, easily accessible i n the Department of Nuclear Medicine, and a 7 emitter at an energy of 140keV which is ideal for the efficiency of the NaI[Tl] detector crystal. T h e heart insert of the thorax phantom was filled with 18.84MBq of calibrated activity at the initial time t. 0  T h e body of the phantom (excluding lung and spinal cord inserts) was filled with  177'.bMBq of activity at time t . T h e Siemens e.cam (dual-head) camera was used for 0  acquisitions, along w i t h low energy, high resolution collimators. Each scan (both planar and S P E C T ) was taken with the presence of a calibration source (contained i n a syringe) with an activity of 10A6MBq at time t . 0  5.1.1  ACQUISITION  PROTOCOLS  Five planar scans were acquired, each over a period of 5 minutes.  T h i s time interval  was intended to resemble that of a patient whole body scan which acquires data over a period of 4.88 minutes i n any particular region. Because the phantom dimensions easily fit within the field of view of the detector, a static planar scan rather than a moving whole-body scan was used.  Simultaneous anterior/posterior scans were acquired with  the phantom and calibration source i n the field of view of the detectors.  Figure 5.1  shows the phantom and detector orientation used i n a l l the planar scans. T h e energy window used to accept/reject photons was a 20% window centered at 140keV - meaning that photons i n the range of 126-154keV as detected by the detector crystal were counted as emission data. T h e data from each detector was acquired i n a 128 x 128 matrix for later analysis. The S P E C T scans were acquired with the two detector heads i n a 90° configuration with respect to one another.  T w o sets of M L A Gadolinium-153 transmission sources  Chapter 5. Phantom Experiments  52  Figure 5.1: P l a n a r imaging of a phantom by two detector heads i n a 180° configuration were placed opposite each detector head so that the transmission data could be collected simultaneous with the emission data.  T h e S P E C T detector and transmission source  orientation is represented i n Figure 5.2 as used for a l l S P E C T scans. T h e scan covered a 180° rotation about the phantom (90° for each detector head) w i t h 64 projections acquired (again, 32 per head). D a t a collection for each projection lasted 30 seconds and was binned into a 128 x 128 matrix. T h e accepted photons were collected from four possible energy windows, rather than one window as was used i n planar studies. T h e first was a 15% window centered about 140keV (emission window), the second a 20% window centered about lOOkeV (transmission window), and the t h i r d and fourth are  Chapter 5. Phantom Experiments  53  scatter windows on both sides of the emission and transmission windows. T h e presence of these scatter windows enables a scatter correction to be performed on the projections, if so desired. T h e t h i r d scatter window allows a correction for the Technetium photons that scatter and result i n being collected into the lower energy G a d o l i n i u m window. T h e transmission window enables attenuation correction to be performed on the data.  Figure 5.2: S P E C T imaging of a phantom by two detector heads i n a 90° configuration  5.1.2  SCAN  SCHEDULE  Phantom activities and scan times were normalized so that the first scan time represented time t = 0 (true activities were decay corrected accordingly). Scans 1, 2, 3, 5 and 7 were  Chapter 5. Phantom Experiments  54  planar scans acquired at t = 0, 87, 214, 352, and 393 minutes and intervening scans 4 and 6 were S P E C T scans acquired at t = 249 and 364 minutes.  5.1.3  R E G I O N O F I N T E R E S T (ROI)  ANALYSIS  In the planar anterior/posterior projections, elliptical R O I ' s (2-dimensional) were drawn around the heart and calibration source regions to obtain the anterior/posterior counts i n the regions. T h e anterior and posterior counts were then averaged by a geometric mean and, for the heart region, corrected for attenuation by using an effective p of 0 . 1 2 c m . -1  The averaged counts i n the calibration R O I , when divided by the known activity i n the calibration source (at the time of the scan), yielded a conversion factor from counts to activity i n MBq. T h i s conversion factor is used to convert the attenuation corrected counts i n the heart region into an activity i n MBq. Once this procedure is carried out for all planar scans, a time-activity curve could be plotted for the heart region. Table 5.1 contains the planar attenuation corrected activities for the heart R O I (and their subsequent errors from the true activity).  Table 5.1: Heart data from planar scans (Experiment # 1 ) .  Scan Number 1 2 3 5 7  Measured Activity (MBq)  True Activity (MBq)  Percent Error  56.55 47.83 38.56 28.93 26.63  17.45 14.77 11.57 8.88 8.21  224 224 233 226 224  (%)  The curve fit to the planar attenuation corrected data yields an equation to represent the activity i n the heart as a function- of time:  Chapter 5. Phantom Experiments  55  Heart Time-Activity Curve (Experiment #1) 60  +  0 X  •  Planar Atten. Corrected Data SPECT Constraints Planar w/ SPECT Constraint True Activity  40 cr CO  >30 > o <  100  200 300 400 Time Post-Injection (minutes)  500  600  Figure 5.3: Heart T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 1  A(t) = 56.81 • e -0.0019-t Figure 5.3 displays the curve fit to the planar attenuation corrected data (+). T h e cumulated activity is determined by integration:  A = r 56.81 • e-°Jo  56.81 0.0019  0019  -*dt =  L  [ - ^ e 0.0019  = 29900MBq • min  -0.0019-tloo .0  Chapter 5. Phantom Experiments  56  thus A = 1.794 x 10 MBq • 5 6  Likewise for the true activity (also represented i n Figure 5.3 (•)):  A(t) = 17.45 • e- 0 . 0 0 1 9 - t  A — 5.511 x  10 MBq-s 5  (1.794 x 10 - 5.511 x 10 ) 6  %Error =  5  5.511 x 10  5  x 100 = 225%  The cumulated activity using the planar attenuation corrected data gives an estimate that is 225% larger than the truth.  5.1.4  SPECT  CONSTRAINT  S P E C T scans were reconstructed using an iterative technique developed by Siemens known as Iterative W. T h e images were reconstructed using 10 iterations and a B u t terworth filter w i t h a cut-off of 0.35. S P E C T scans 4 and 6, acquired at t = 2 4 9 m i n and 3 6 4 m m , were used to constrain the planar data. F r o m the 3-dimensional iteratively reconstructed image, and using the Display software to determine the counts i n the heart and calibration R O I ' s , the activity of the heart was determined as l l . 8 4 M . B g at t = 2 4 9 m m and 9.27MBq at t = 364mm. Table 5.2 illustrates the difference between the S P E C T values and the truth, giving an idea of how well S P E C T performs quantitatively. The planar curve, at t — 2 4 9 m m , produces an activity of  -0.0019-249  A(t = 249) = 56.81 • e  =  35A0MBq  Chapter 5. Phantom Experiments  57  Table 5.2: Heart data from S P E C T scans (Experiment # 1).  Scan Number  Measured Activity (MBq)  True Activity (MBq)  4 6  11.84 9.27  10.87 8.74  Likewise for t = 3 6 4 m m , A(t — 364) = 28A5MBq.  Percent Error (%) 9 6  T h e ratio of the planar value (at  t — 2 4 9 m m and 364mm) to the S P E C T value (at t = 2 4 9 m m and 364mm) yields the normalization factors used to constrain the planar data.  Ni = 1  35.40MBg  n  11.8AM Bq  n  n  = 2.99  28A5MBg N  -  l  9.27MBq ~  '°  3  7  A normalization factor Nj can be determined for each S P E C T scan j that is acquired. A n average N is taken of all the normalization factors then a l l planar data is divided by the mean normalization factor to achieve the S P E C T constrained data.  ^N N =  1±  1  =  2  2-99 + 3.07 2  Table 5.3 displays the results of dividing a l l of the planar activities by the normalization factor. The S P E C T constraint yields a new curve fit to the d a t a (represented i n Figure 5.3  A(t) = 18.80 -e -0.0019-t A n d the cumulated activity is:  Chapter 5. Phantom  Experiments  58  Table 5.3: Planar data re-normalized using the S P E C T Constraint for the Heart R O I (Experiment # 1).  Scan Number  Constrained Activity (MBq)  True Activity (MBq)  1 2 3 5 7  18.66 15.79 12.73 9.55 8.79  17.45 14.77 11.57 8.88 8.21  A = 5.937 x 10 MBq5  Percent Error (%) 7 7 10 8 7  s  The S P E C T constraint has brought the cumulated activity estimate to w i t h i n 8% of the true cumulated activity within the heart region of the phantom.  5.2  TECHNETIUM-99M PHANTOM EXPERIMENT #  2  The same thorax phantom was used for this experiment with the addition of two spherical tumor inserts. One tumor (2.54cm diameter) insert was placed under the right lung and the other (1.86cm diameter) was placed on the lateral surface of the left lung. T h e heart was injected w i t h an activity of 20 AM Bq, the body with 127.2MBq, (Tumor # 1) w i t h 8.87MBq,  the right-lung tumor  and the left-lung tumor (Tumor # 2) w i t h 8.99MBq.  calibration source used i n the planar scans had an initial activity of 9A\MBq  The  and an  additional calibration source (on the surface of the phantom) was used in the S P E C T scans ( 1 0 . 2 1 M 5 g ) i n order to test the effects of attenuation on the calibration source. A l l of the quoted activities are decay corrected so that the first scan coincides w i t h time  Chapter 5. Phantom Experiments  t. 0  59  Figure 5.4 illustrates the thorax phantom and the two tumor inserts used i n the  experiment.  Figure 5.4: T h o r a x phantom with two spherical tumor inserts - one underneath the right lung and one on the lateral side of the left lung.  5.2.1  ACQUISITION  PROTOCOLS  The planar and S P E C T acquisition protocols are identical to those described i n Technetium99m Phantom Experiment #1.  5.2.2  SCAN SCHEDULE  Scans 1, 2, 4. 6 and 7 were planar scans acquired at t = 0 , 9 2 , 2 1 5 , 3 5 3 , and 3 8 1 minutes and scans 3 and 5 were S P E C T scans acquired at t = 103 and 3 2 6 minutes.  Chapter 5. Phantom Experiments  5.2.3  60  REGION OF INTEREST (ROI) ANALYSIS  Region of interest analysis was performed as described i n the previous experiment. A g a i n the planar attenuation correction was performed using an effective u. of 0 . 1 2 c m . Tables - 1  5.4, 5.5, and 5.6 contain the planar attenuation corrected activities for the heart, tumor 1, and tumor 2, respectively. Table 5.4: Heart data from planar scans (Experiment # 2).  Scan Number  Measured Activity (MBq)  True Activity (MBq)  1 2 4 6 7  46.11 39.22 31.51 24.13 22.84  20.4 17.1 13.6 10.4 9.89  Percent Error (%) 126 129 132 132 131  Table 5.5: Tumor 1 data from planar scans (Experiment #2).  Scan Number 1 2 4 6 7  Measured Activity (MBq)  True Activity (MBq)  18.21 15.39 12.12 9.41 8.92  8.87 7.45 5.90 4.54 4.30  Percent Error (%) 105 107 105 107 107  HEART  The curve fit to the planar attenuation corrected data yields the following:  Chapter 5. Phantom Experiments  61  Table 5.6: Tumor 2 data from planar scans (Experiment # 2).  Scan Number  Measured Activity  1 2 4 6 7  (MBq)  True Activity (MBq)  23.96 20.11 16.02 12.11 11.80  8.99 7.55 5.98 4.60 4.36  A(t) = 46.40 • e -  0  0 0 1 8  Percent Error (%) 167 166 168 163 171  '*  Figure 5.5 displays the curve fit to the planar attenuation corrected data (+). F r o m this fit, the cumulated activity is A = 1.547 x 10 MBq • s. 6  The true activity (also represented i n Figure 5.5(-)) and the resulting cumulated activity are: A{t) = 20.4 • e - ° -  o o m  A = 6.442 x 10 MBq • s 5  T h e cumulated activity using the planar attenuation corrected data gives an estimate that is 140% larger than the truth.  TUMOR  1  The curve fit to the planar attenuation corrected data yields the following:  A{t) = 18.2 • e -  0  0 0 1 9  '  4  Chapter 5. Phantom Experiments  62  Heart Time-Activity Curve (Experiment#2)  Ql  0  1  100  1  1  1  200 300 400 Time Post-Injection (minutes)  I  I  500  600  Figure 5.5: Heart T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 2. Figure 5.6 displays the curve fit to the planar attenuation corrected data (+). From this fit, the cumulated activity is A = 5.747 x 10 MBq • s. 5  The true activity (also represented i n Figure 5.6(-)) and the resulting cumulated activity are: A{t) = 8.87 • -  o  m  i  9  t  e  A = 2.801 x 10 MBq • s 5  The cumulated activity using the planar attenuation corrected data gives an estimate that is 105% larger than the truth.  Chapter 5. Phantom Experiments  63  Tumor 1 Time-Activity Curve (Experiment#2)  QI 0  1  100  1  1  i  200 300 400 Time Post-Injection (minutes)  i  500  I 600  Figure 5.6: T u m o r 1 T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 2, TUMOR  2  The curve fit to the planar attenuation corrected data yields the following: A(t) = 24.0 • - ° -  0 0 1 ! H  e  Figure 5.7 displays the curve fit to the planar attenuation corrected data (+). F r o m this fit, the cumulated activity is A = 7.579 x l0 MBq 5  • s.  The true activity (also represented i n Figure 5.7(-)) and the resulting cumulated activity are: A(t) = 8.99 • e - ° -  o o m  Chapter 5. Phantom Experiments  64  Tumor 2 Time-Activity Curve (Experiment#2)  200 300 400 Time Post-Injection (minutes)  Figure 5.7: T u m o r 2 T i m e - A c t i v i t y curve for the thorax phantom i n Experiment # 2.  A = 2.839 x 10 MBq • s 5  The cumulated activity using the planar attenuation corrected data gives an estimate that is 167% larger than the truth.  5.2.4  SPECT  CONSTRAINT  Image reconstruction using the Iterative W method resulted i n strong artifacts i n the image, likely due to the high activities i n the tumors. For this reason the S P E C T scans  Chapter 5. Phantom  65  Experiments  were reconstructed using another iterative technique known as OSEM Expectation  Maximization.  or Ordered Subset  The images were reconstructed over 15 iterations using a  Butterworth filter w i t h a cut-off of 0.35. T h e OSEM  method returned a cleaner image  with less artifacts. Scans 3 and 5, acquired at t = 103mm and 3 2 6 m m , were S P E C T scans and were used to constrain the planar data.  F r o m the 3-dimensional iteratively  reconstructed image, the Display software determined the counts i n the heart, tumor 1, tumor 2, and calibration R O I ' s . Tables 5.7, 5.9 and 5.11 illustrate the difference of the S P E C T values from the truth, demonstrating how S P E C T performs quantitatively.  HEART  Table 5.7: Heart data from S P E C T scans (Experiment # 2).  Scan Number 3 5  Measured Activity (MBq)  True Activity (MBq)  19.02  16.77  10.10  10.98  Percent Error (%) 13 -8  A t times t = 103mm and 3 2 6 m m , the planar curve yields activities of and 25.80MBq  respectively. Determining the normalization factors:  = ~ = 203 19.02 and iV = ™ ° = 2.55 10.10 2 2  2.03 + 2.55 N = = 2.29 n  n  n  38.55MBq  Chapter 5. Phantom Experiments  66  Table 5.8: Planar data re-normalized using the S P E C T Constraint for the Heart R O I (Experiment # 2).  Scan Number  Constrained Activity (MBq)  True Activity (MBq)  Percent Error  1 2 4 6 7  20.14 17.13 13.76 10.54 9.97  20.4 17.13 13.56 10.43 9.89  -1 0 1 1 1  (%)  The results of using the above normalization factor to constrain the planar data are summarized i n Table 5.8. The new fit to the S P E C T constrained data (represented i n Figure 5.5 (*)) is: A(t) = 21.16 • - ° -  0 0 2 ( H  e  giving a cumulated activity A of 6.348 x 10 MBq-s which is —1% from the true cumulated 5  activity.  TUMOR  1  Table 5.9: Tumor 1 data from S P E C T scans (Experiment # 2).  Scan Number  Measured Activity (MBq)  True Activity (MBq)  Percent Error  3 5  6.35 5.07  7.29 4.77  -13 6  (%)  A t times t = 1 0 3 m m and 326mm, the planar curve yields activities of \A.97MBq and 9.80MBq respectively. Determining the normalization factors:  67  Chapter 5. Phantom Experiments  14 97  and N , - m = 1.93 5.07  ^  =  2.36+ 1.93  =  2  _  1 5  The results of using this normalization factor to constrain the planar data are summarized i n Table 5.10. Table 5.10: Planar data re-normalized using the S P E C T Constraint for the Tumor 1 R O I (Experiment # 2).  Scan Number 1 2 4 6 7  Constrained Activity (MBq)  True Activity (MBq)  8.47 7.16 5.64  8.87 7.45 5.90 4.54 4.30  4.38 4.15  Percent Error (%) -5 -4 -4 -4 -3  The new fit to the S P E C T constrained data (represented i n Figure 5.6 (*)) is:  A(t) = 8.20 • e"  0 0 0 1 7  '  4  giving a cumulated activity A of 2.894 x 10 MBq- s which is 3% from the true cumulated 5  activity.  Chapter  5.  Phantom  68  Experiments  Table 5.11: Tumor 2 data from S P E C T scans (Experiment # 2).  TUMOR  Scan Number  Measured Activity (MBq)  True Activity (MBq)  3 5  9.18 7.32  7.39 4.84  Percent Error (%) 24 51  2  A t times t = 1 0 3 m m and 326mm, the planar curve yields activities of 1 9 . 7 3 M i ? g and 12.92MBq  respectively. Determining the normalization factors:  and 12.92  *  = TJ2  =  1 7 7  2 The results of using this normalization factor to constrain the planar data are summarized in Table 5.12. The new fit to the S P E C T constrained data (represented in Figure 5.7 (*)) is:  •A(t) = 11.82 • e -0.0017-t giving a cumulated activity A of 4.172 x 10 MBq-s 5  activity.  which is 47% from the true cumulated  Chapter 5. Phantom Experiments  69  Table 5.12: Planar data re-normalized using the S P E C T Constraint for the Tumor 2 R O I (Experiment # 2 ) .  5.3  Scan Number  Constrained Activity (MBq)  True Activity (MBq)  Percent Error  1 2 4 6 7  12.22 10.26 8.17 6.18 6.02  8.99 7.55 5.98 4.60 4.36  36 36 37 34 38  (%)  DISCUSSION  Results from the planar studies compiled i n Tables 5.1, 5.4, 5.5 and 5.6, and the S P E C T studies compiled i n Tables 5.2, 5.7, 5.9 and 5.11 demonstrate how planar and S P E C T data compare quantitatively once corrected for attenuation. In experiment # 1 , by using a S P E C T constraint, the cumulated activity i n the heart improved from a 225% overestimate of the t r u t h (planar attenuation correction) to an 8% overestimate of the truth using the S P E C T attenuation correction. Likewise, i n experiment #2, the heart A i m proved from 140% error to —1% error, the tumor #1 from 105% error to 3% error, and the tumor #2 from 167% error to 47% error - all the result of using a S P E C T constraint rather than planar data alone. A l t h o u g h tumor #2 experienced a drastic improvement in the estimate of A, the resulting error of 47% is potentially due to persistent artifacts in the image around tumor #2. Tumor #2 was more susceptible to reconstruction artifacts due to its high activity contained i n a small volume, along w i t h its placement near the edge of the phantom. The method selected to reconstruct the S P E C T data should not be carelessly chosen. Improper reconstruction w i l l result i n image artifacts that w i l l carry over to the 3-D R O I  Chapter 5. Phantom Experiments  70  software thus affecting the activity estimates i n the source regions. It is wise to be most cautious when dealing with regions of high activity, as these regions are most prone to reconstruction artifacts. For the levels of activity used i n Experiment # 1 , reconstruction by Iterative W d i d not produce any noticeable artifacts. However, at the high tumor activities used i n Experiment # 2 , Iterative W did produce noticeable artifacts, prompting the decision to t r y another reconstruction method. OSEM improved the image quality sufficiently to consider it a worthy option when dealing w i t h high activities.  5.4  O T H E R FACTORS AFFECTING  QUANTITATION  Some factors are often overlooked when considering quantitation i n imaging. Camera systems are generally designed to cope with matters affecting image quality in diagnostic applications but quantitation may break down when most needed - at the high levels of activity necessary for therapeutic applications. T h e two factors considered here, namely P i x e l Saturation and Dead-Time, were investigated as to their effect on image quantitation.  5.4.1  PIXEL  SATURATION  W h e n the computer stores photon counts into the specified matrix, each pixel in that matrix is designated to hold a specific data type. In most cases this data type is unsigned integer, also known as unsigned short integer. T h i s data type allows for numbers 0 to 65535  to represent the counts collected i n that particular pixel. T h e purpose of assigning  a specific data type to hold the data is for the sake of memory storage requirements. A n unsigned integer only requires 2 bytes of memory space for storage whereas an unsigned long integer (from 0 to 4 , 2 9 4 , 9 6 7 , 2 9 5 ) requires 4 bytes of memory space for storage. Saving all image matrices with a data type of unsigned long integer would seem excessive  Chapter 5. Phantom Experiments  71  and a waste of memory since most pixel counts would rarely exceed 65535. T h i s is the valid justification of why most imaging software w i l l store the data as unsigned integer. This justification is perfectly warranted for most needs of a Nuclear Medicine department as most applications are diagnostic i n nature. B u t w i t h the increasing interest i n therapeutic applications, the justification may no longer be valid. T o investigate potential pixel saturation, an experiment was performed w i t h a syringe initially containing 16.34M-Bg of T c - 9 9 m , scanned over a period of 5 minutes (simulating a whole body planar scan), and the data was acquired i n a 64 x 64 matrix. Analysis of the image proved that quantitation was lost as some of the pixels were saturating (ie. no longer acquiring counts) after they reached a count of 65535. Once pixels stop acquiring data, when data is still present to be counted, accurate quantitation of the data is no longer possible. Another experiment was performed using a syringe of 13.23MBq of T c - 9 9 m , scanned over a period of 1 minute, and the d a t a was acquired i n a 256 x 256 matrix. It was concluded that no pixel saturation was present i n this case. T h e effects of pixel saturation can be overcome by increasing the size of the acquisition matrix, reducing the acquisition time, decreasing the activity or increasing the size of the activity source (same activity but i n a larger volume). W h e n considering therapy, some of these factors are controllable and others are beyond our control.  5.4.2  DEAD-TIME  When counts (scintillation events) are collected by the gamma camera at rapid acquisition rates, the gamma camera may fail to detect some scintillation events that interact w i t h the detector crystal. T h e loss of counts is due to the dead-time of the electronics of the detection and computer instrumentation. Dead-time is defined as the period of time after acquisition of a scintillation during which the gamma camera and computer electronics  Chapter 5. Phantom Experiments  72  are unable to respond to another scintillation [38]. W h e n the source activity is increased (increasing the scintillation events) the number of counts lost w i l l increase as well. In regions of high count rates, dead-time results i n a loss of quantitation of the data. T h e effects of dead-time w i l l be more pronounced i n the case of high levels of activity. Two experiments were performed to investigate the effects of dead-time.  T h e first  experiment coincided w i t h one of the pixel-saturation experiments described above. T h e experiment was performed using a syringe of 13.23M Bq of Tc-99m, scanned over a period of 1 minute, and the data was acquired i n a 256 x 256 matrix. Six scans were acquired over a period of 5 hours w i t h the scan times accurately recorded. T h e total counts i n each of scans 2 through 6 were corrected for physical decay i n order to normalize each scan to Scan # 1. Performing this normalization allows us to see any effects of dead-time since a horizontal plot (Counts vs. Scan) is free of significant dead-time whereas a plot that deviates from the horizontal reveals the influence of dead-time. T h e results for this experiment are plotted i n Figure 5.8. Taking statistical errors into account the plot i n Figure 5.8 proves to be a horizontal line, suggesting that there is no significant dead-time present. In other words, quantitation is not lost by potential dead-time effects. To explore this further, another experiment was performed - this time using both detectors 1 and 2 of the Siemens M u l t i S P E C T 2 (MS2)  and detector 2 of the Siemens e.cam. T o simulate an extended source, like the  area of a kidney, a petri dish of diameter 8.88cm (61.9cm area) was used. A n activity 2  of 347'MBq was distributed over the bottom of the petri dish w i t h m i n i m a l liquid (to minimize attenuation).  Four 5 minute scans were acquired (in 128 x 128 matrices) on  both the e.cam and the M S 2 over a period of 5 hours. T h e first scan on the e.cam was plagued by pixel saturation, thus the e.cam scan time was reduced to 3 minutes. T h e  Chapter 5. Phantom  x 10  Experiments  73  Dead-Time Investigation for Siemens E.Cam Detector 1  Scan Number  Figure 5 . 8 : Results of the D e a d - T i m e investigation for detector 1 of the Siemens e. cam camera. M S 2 data d i d not have any pixel saturation, likely due to the fact that the M S 2 camera had ultra-high resolution collimators w i t h less sensitivity than the high resolution collimators used on the e.cam. After a l l scans were normalized (decay corrected) to the first scan, they were plotted on a Counts vs. Scan plot. Results are summarized i n Figure 5 . 9 . T a k i n g statistical errors into account, a l l three plots display a non-horizontal trend. T h e e.cam detector 2 is most noticeably impacted by the effects of dead-time.  T h i s is likely due to the  collimator difference between the e.cam and the M S 2 . Quantitation w i l l be restored when the activity has decayed to a manageable level. These results demonstrate a necessary concern over dead-time i n image quantitation. A l t h o u g h this experiment may prove to be an extreme case (ie. high activity over a small region w i t h little attenuation), it also  Chapter 5. Phantom Experiments  74  Dead-Time Investigation for Siemens E.Cam and M S 2  x 10  0 + x  4.8  E.Cam detector 2 M S 2 detector 1 M S 2 detector 2  4.6  £ 4.2  o Z  3.8  3.6 1.5  2  2.5  3.5  Scan Number  Figure 5.9: Results of the Dead-Time investigation for detector 2 of the Siemens e.cam and detectors 1 and 2 of the Siemens M u l t i S P E C T 2 . proves that dead-time cannot be ignored as a factor affecting quantitation - especially at therapeutic levels of activity. W h e n quantitation is important, as it is i n dosimetry, protocols need to be established for particular cameras regarding dead-time limitations (ie. activities, m a t r i x sizes, collimators, scan times), or dead-time needs to be measured and accounted for.  CHAPTER 6  CONCLUSIONS  This thesis proposes a method for Internal Dosimetry that implements patient-specific S P E C T scans that are corrected for attenuation. To conclude this thesis, I w i l l briefly summarize the work that has been presented and suggest an extension of this work for clinical applications.  6.1  SUMMARY OF T H E W O R K  In order to form an understanding of dosimetry, it is fundamental to be familiar with radiations relevant to Nuclear Medicine, how these radiations interact w i t h tissue, and how it is possible to detect where radiation originates from w i t h i n a patient.  The in-  troductory chapter built the foundation of this understanding by introducing emissions involving a and j3 particles and 7 photons - discussing how these particles can be absorbed or scattered in matter - and how the gamma camera is used to detect the 7 emissions emanating from a patient. Chapter 2 went on to describe how the effect of radiation on tissue must be quantified by estimating the internal absorbed dose. T h i s is the goal of Internal Dosimetry. The primary method to estimate the absorbed dose is known as the M I R D Protocol. 75  A  Chapter 6.  Conclusions  76  comprehension of the M I R D protocol is v i t a l as a l l current methods have derived from it in one form or another. The M I R D protocol explains how the absorbed dose is calculated goes on to detail how the absorbed dose is measured.  but Chapter 3  T o do this, the two primary  modalities used to perform measurements are explained, namely planar and S P E C T imaging. In order to quantify planar and S P E C T data, a correction for attenuation must be performed. T h i s chapter outlines current methods in dosimetry that arose from the need to improve absorbed dose estimates, hence the proposal of this thesis is included here.  B y keeping most scans planar, yet implementing at least 2 S P E C T scans, we  have gained the benefits of 3-D information, not to mention a promising technique for attenuation correction. I have put forth a method to provide higher accuracy than planar methods alone. Simulations were performed to give an idea of the errors incurred by planar methods of attenuation correction. T h e high errors that stem from planar attenuation correction justify the motivation to improve this favored method somehow. For averaging anterior and posterior image R O I ' s , it was detemined that a geometric mean demonstrated more quantitative accuracy than an arithmetic mean - suggesting that i f planar methods are the sole method available, a geometric mean should be used when processing the data. P h a n t o m experiments were carried out, not only to describe i n detail the protocol presented here, but also to illustrate the results of this protocol. Estimates of the cumulated activity by using a S P E C T constraint demonstrates that the method shows much promise.  A good method of attenuation correction (the M L A transmission source) is  crucial to this method, not to mention an appropriate reconstruction method. For all regions, the cumulated activity estimates improved from an average error of 159% to an average error of 14% when using a S P E C T constraint over planar exclusively. These  Chapter 6.  Conclusions  77  results are especially promising i n light of the need for improvements in dosimetry for therapeutic applications. To date, quantitative planar methods have maintained their popularity by their simplicity and clinically practical time requirements. Never has there been as much of a push to improve planar estimates as now, w i t h the promise of new therapeutic possibilities looming on the horizon. W i t h the high levels of activity required for therapy, it is no longer sufficient to settle for dose estimates that exceed 100% error. A t this level of uncertainty, either patient's lives could be at risk or adequate therapy may not be provided. Implementing a dosimetry protocol that maintains the majority of scans as planar yet incorporates regional S P E C T scans would reduce the need to re-train staff on a completely new protocol while also producing more accurate dose estimates. If patients can be successfully treated with new radiopharmaceuticals, the extra clinical time needed to perform two or more regional S P E C T scans should be considered worthwhile. Other factors that are often taken for granted but may affect quantitation have also been covered, namely pixel saturation and dead-time. Quantitation of the data is lost if pixels begin to saturate at high levels of activity. P i x e l saturation does not pose much of a problem for the short S P E C T acquisitions but planar scans can prove vulnerable to saturation over a 4 . 8 8 m m whole-body acquisition. Dead-time effects are potentially significant at therapeutic levels of activity but quantitation may be restored if these effects are measured and accounted for.  6.2  SUGGESTIONS FOR CLINICAL  APPLICATIONS  A l t h o u g h these results may be impressive for a phantom study, reality dictates that a proposed method must be practical for use on patients.  To accurately produce time-  activity curves, S P E C T scans would have to be performed over any region containing  Chapter 6.  Conclusions  78  significant uptakes of activity. T h i s may or may not prove to be time consuming for some radiopharmaceutical distributions. Ideally, this potential setback could be overcome by using a transmission source on a camera that is capable of performing helical S P E C T . Instead of a regional S P E C T scan, a whole-body S P E C T scan could be performed. A g a i n , w i t h the aid of a good reconstruction method, a l l significant regions i n the whole-body S P E C T image could be used to constrain the corresponding planar images. The development of a superior reconstruction method that is capable of producing high-activity images free of artifacts would also be a worthwhile pursuit.  A n effective  means of attenuation correction i n S P E C T (perhaps transmission helical S P E C T ) , along with improved reconstruction methods, would certainly help pave the way for a growing interest in therapeutic applications in Nuclear Medicine.  BIBLIOGRAPHY Darrell R . Fisher. R a d i a t i o n dosimetry for radioimmunotherapy: current capabilities and limitations. Cancer, 73(3):905-911, 1994.  an overview of  M . E . N o z and G . Q . Maguire Jr. Radiation Protection in the Radiologic and Health Sciences. L e a and Febinger, Philadelphia, 1979. L . T . D i l l m a n and F . C . Vonder Lage. Radionuclide decay schemes and nuclear parameters for use in radiation dose estimation. MIRD Pamphlet No. 10. T h e Society of Nuclear Medicine, New York, 1975. J . A . Sorenson and M . E . Phelps. Physics in Nuclear Medicine. W . B . Sauders C o m pany, 1987. B a r b a r a Y . Croft. Single-Photon Emission Computed Tomography. Year B o o k M e d ical Publishers, Inc., Chicago, 1986. R . W . Howell, B . W . Wessels, and R . Loevinger. T h e M I R D perspective 1999. J. Nucl. Med., 40:3S-10S, 1999. Medical Internal R a d i a t i o n Dose Committee. Med., 29(Suppl.l), 1968.  M I R D Pamphlet Nos. 1-3. J. Nuc.  R. Loevinger and M . Berman. MIRD Pamphlet No.l Revised: A revised schema for calculating the absorbed dose from biologically distributed radionuclides. T h e Society of Nuclear Medicine, New York, 1976. R. Loevinger, T . F . Budinger, and E . E . Watson. MIRD Primer for absorbed dose calculations. Revised Edition. T h e Society of Nuclear Medicine, New York, 1991. G . L . Brownell, W . H . Ellett, and A . R . Reddy. Absorbed fractions for photon dosimetry. M I R D Pamphlet no.3. J. Nuc. Med., 9(Suppl.l):27-39, 1968. W . S . Snyder, M . R . Ford, and G . G . Warner et a l . 'S' absorbed dose per unit cumulated activity for selected radionuclides and organs. MIRD Pamphlet No. 11. T h e Society of Nuclear Medicine, New York, 1975. L . T . D i l l m a n . Radionuclide decay schemes and nuclear parameters for use i n radiation dose estimation, P a r t 2. M I R D Pamphlet No.6. J. Nucl. Med., l l ( S u p p l . 4 ) : 5 - 3 2 , 1970. L . T . D i l l m a n . Radionuclide decay schemes and nuclear parameters for use i n radiation dose estimates. M I R D Pamphlet No.4. J. Nucl. Med., 10(Suppl.2):20, 1969. 79  Bibliography  80  [14] M . Berman. Kinetic models for absorbed dose calculations. MIRD Pamphlet No.12. T h e Society of Nuclear Medicine, New York, 1977. [15] W . S . Snyder, M . R . Ford, and G . G . Warner. Estimates of specific absorbed fractions for photon sources uniformly distributed in various organs of a heterogeneous phantom. MIRD Pamphlet No. 5. T h e Society of Nuclear Medicine, New York, 1990. [16] Schlomo Hoory. Internal dosimetry evaluation: a computerized approach. Health Phys., 50(2):195-201, 1986. [17] Peter J . Mountford. Internal dosimetry: Developments and limitations. Eur. J. Nuc. Med., 23:491-493, 1996. [18] G o p a l B . Saha. Physics and Radiobiology of Nuclear Medicine. Springer-Verlag, New York, 1993. [19] W . S . Snyder, M . R . Ford, and G . G . Warner et al. Estimates of absorbed fractions for monoenergetic photon sources uniformly distributed i n various organs of a heterogeneous phantom. M I R D Pamphlet No.5. J. Nucl. Med., 10(Suppl.3):5-52, 1969. [20] M . J . Berger. Energy deposition i n water by photons from point isotropic sources. M I R D Pamphlet No.2. J. Nuvl. Med., 9(Suppl.l):15-25, 1968. [21] M . Cristy and K . F . Eckerman. Specific absorbed fraction of energy at various ages from internal photon source. Vols. I-VII. O a k Ridge N a t i o n a l Laboratory, O a k Ridge, T N , 1987. [22] R . J . Cloutier, S . A . S m i t h , and E . E . Watson et al. Dose to the fetus from radionuclides i n the bladder. Health Phys., 25:147-161, 1973. [23] T . K . Johnson. M A B D O S : a generalized program for internal radionuclide dosimetry. Computer Methods and Programs in Biomedicine, 27:159-167, 1988. [24] M . Tagesson, M . Ljungberg, and S.E. Strand. Transformation of activity distribution i n quantitative spect to absorbed dose distribution i n a radionuclide treatment planning system. J. Nuc. Med., page 123P, 1994. [25] E l i E . Furhang. A Monte C a r l o approach to patient-specific dosimetry. Med. Phys., 23(9):1523-1529, Sept.1996. [26] G.Sgouros. Three-dimensional dosimetry for radioimmunotherapy treatment planning. J. Nucl. Med., 34:1595-1601, 1993. [27] G.Sgouros, G . Barest, and J . T h e k k u m t h a l a et al. Treatment planning for internal radionuclide therapy: Three-dimensional dosimetry for nonuniformly distributed radionuclides. J. Nucl. Med., 31:1884-1891, 1990.  Bibliography  81  [28] H . B . G i a p , D . J . Macey, J . E . B a y m o u t h , and A . L . Boyer. Validation of a dose-point kernel convolution technique for internal dosimetry. Phys. Med. Biol, 40:365-381, 1995. [29] H . B . G i a p , D . J . Macey, and D . A . Podoloff. Development of a S P E C T - b a s e d threedimensional treatment planning system for radioimmunotherapy. J. Nucl. Med., 36:1885-1894, 1995. [30] A . T . S.-Stelson, E . E . Watson, and R . J . Cloutier. A hostory of medical internal dosimetry. Health Phys., 69(5):766-782, 1995. [31] P . B . Zanzonico, R . E . Bigler, G . Sgouros, and A . Strauss. radiation dosimetry. Semin. Nucl. Med., 19:47-61, 1989.  Quantitative S P E C T in  [32] K . K o r a l , K . R . Zasadny, and M . L . Kessler et al. A method using C T - S P E C T fusion plus conjugate views for dosimetry i n I-131-MoAb therapy of l y m p h o m a patients. J. Nuc. Med., 35:1714-1720, 1994. [33] S . A . Larsson. Gamma Camera Emission Tomography: development and properties of a multi-sectional emission computed tomography system. P h D thesis, University of Stockholm, Sweden, 1980. [34] J . A . Siegel, S.R. Thomas, and J . B . Stubbs et a l . M I R D Pamphlet N o . 16: Techniques for quantitative radiopharmaceutical biodistribution data acquisition and analysis for use i n human radiation dose estimates. J. Nucl. Med., 40:37S-61S, 1999. [35] L . T . Chang. A method for attenuation correction i n radionuclide computed tomography. IEEE Trans. Nuc. Set., NS-25:638-643, 1978. [36] A . Celler, A . Sitek, E . Stoub, P . H a w m a n , R . Harrop, and D . Lyster. Investigation of an array of multiple line sources for S P E C T transmission scans: Simulation, phantom and patient studies. J. Nuc. Med., 39(12), 1998. [37] C . J . D y k s t r a , R . Harrop, A . M . Celler, and M . S . A t k i n s . Quantitative 3 D data extraction using contiguous volumes. IEEE Trans. Nuc. Sci., 44(4):1571-1577, 1997. [38] M . J . Gelfand and S.R. Thomas. Effective Use of Computers in Nuclear Medicine. M c G r a w - H i l l , New York, 1988.  

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.831.1-0085101/manifest

Comment

Related Items