Citation
The application of cadmium telluride as in hole semiconductor radiation detectors in an X Y matrix radioisotope camera

Material Information

Title:
The application of cadmium telluride as in hole semiconductor radiation detectors in an X Y matrix radioisotope camera
Series Title:
The application of cadmium telluride as in hole semiconductor radiation detectors in an X Y matrix radioisotope camera
Creator:
Allison, Jerry David
Publisher:
Jerry David Allison
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Amplifiers ( jstor )
Collimators ( jstor )
Crystals ( jstor )
Electric potential ( jstor )
Electrodes ( jstor )
Parallel lines ( jstor )
Photons ( jstor )
Radionuclides ( jstor )
Semiconductors ( jstor )
Signals ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
022590341 ( ALEPH )
04718263 ( OCLC )

Downloads

This item has the following downloads:


Full Text














THE APPLICATION OF CADMIUM TELLURIDE AS IN HOLE SEMICONDUCTOR
RADIATION DETECTORS IN AN X Y MATRIX RADIOISOTOPE CAMERA









BY -

JERRY DAVID ALLISON


A DISSERTATION PRESENTED TO
UNIVERSITY
IN PARTIAL FULFILLMENT OF THE
DOCTOR OF


THE GRADUATE COUNCIL OF THE
OF FLORIDA
REQUIREMENTS FOR THE DEGREE OF
PHILOSOPHY


UNIVERSITY OF FLORIDA


1978



































To My Parents














ACKNOWLEDGMENTS


The members of my advisory committee (Drs. W. Emmett Bolch, Valerie

A. Brookeman, Lawrence T. Fitzgerald, Walter Mauderli, Genevieve S.

Roessler and Henri A. Van Rinsvelt) were instrumental in the formula-

tion, accomplishment and documentation of this research. Dr. Mauderli

served as chairman of the committee and provided special assistance

with analog electronics. Dr. Fitzgerald provided special assistance

with digital electronics and microcomputer applications. Dr. Bolch,

through the Department of Environmental Engineering Sciences, provided

a graduate assistantship during my first year at the University.

Dr. Genevieve Roessler counseled me with regard to course work and

helped me find appropriate assistantships throughout my graduate study

at the University.

Dr. Clyde M. Williams, as Chairman of the Department of Radiology,

provided a graduate assistantship for myself and financial support for

my research project.

The person most responsible for making this dissertation possible

is my wife, Jacquie. Jacquie and I left rather secure professional

careers and a very pleasant social environment in Newport News,

Virginia, so that I could pursue a doctorate at the University of Florida.

Not only did Jacquie provide moral support but also supported our family

financially by teaching in the secondary schools of Alachua County.

She bore our second child and completed her master's degree during our

three year stay in Gainesville. Jacquie made this dissertation possible.














TABLE OF CONTENTS


ACKNOWLEDGMENTS . . . . . .

ABSTRACT . . . . . . . .

CHAPTER

1. INTRODUCTION . . . . .

2. HISTORICAL DEVELOPMENT . .

3. CAMERA PROPERTIES . . . .
Advantages of a CdTe Matrix .
Advantages of In Hole Detectors
Disadvantages . . . . .

4. CAMERA DESIGN AND CONSTRUCTION
Detector Matrix . . . .
Collimator . . . . .
Analog Circuitry . . . .
Analog to Digital Circuitry. .
Digital Circuitry . . . .
Microcomputer . . . . .
Videographics Interface . ..
Summary . . . . . .

5. RESULTS . . . . . .

6. DISCUSSION OF RESULTS . . .

7. CLINICAL FEASIBILITY . . .

BIBLIOGRAPHY . . . . . . .

BIOGRAPHICAL SKETCH . . . . .


Page

iii

v










Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy


THE APPLICATION OF CADMIUM TELLURIDE AS IN HOLE SEMICONDUCTOR
RADIATION DETECTORS IN AN X Y MATRIX RADIOISOTOPE CAMERA


By

Jerry David Allison

August 1978


Chairman: Walter Mauderli, D. Sc.
Major Department: Nuclear Engineering Sciences


A radioisotope camera that has superior performance in comparison

to conventional devices, has been designed, constructed, and tested.

Radioisotope cameras are instruments used for imaging the distribution

of radioisotopes in vivo. A camera creates a two dimensional image of

the three dimensional radioisotope distribution within an object. Con-

ventional radioisotope cameras are mainly of the Anger type. The radia-

tion detector in an Anger camera is a sodium iodide scintillation crystal

that is located in back of a collimator. The collimator creates a

spatial relationship between the image and the radioisotope distribution.

The new camera is a small experimental camera that has a matrix of

cadmium telluride semiconductors as the detecting medium. The detectors

are located within the collimating holes (rather than behind them).

Coincident outputs from each row and column are used to digitally

identify the individual detector in which the energy deposition occurred.

Direct digital localization of detected events and the attenuation of










scatter radiation between matrix elements by the collimator septa that

separate detectors, enables improved resolution and imaging time.

The superior radiation detection characteristics of cadmium telluride

(in comparison to sodium iodide) could provide superior discrimination

against scattered radiation that contributes only spurious information

to the image. The availability of large numbers of small cadmium

telluride detectors that have uniform radiation detection properties would

make a clinical camera viable.













CHAPTER 1

INTRODUCTION


Radioactive tracer studies are an integral part of modern medical

diagnostic procedures. By administering radioactive tracers (radio-

pharmaceuticals) that are specific for some physiological function,

physicians are able to determine the level of that function within a

patient in terms of physical and chemical phenomena and hence determine

the site and rate of important biological processes. Radioisotope

cameras are the instruments that enable many of the relatively nonin-

vasive radioactive tracer studies that are currently prevalent in nuclear

medicine. It is because of the fact that improvements in spatial or

temporal resolution of radioisotope cameras have the potential of making

better diagnostic information available to the physician as well as

reducing radiation dose to the patient, that the research described

herein was undertaken and is important.

Radioisotope cameras image the distribution of radioisotopes in

vivo. A camera creates a two dimensional image of the three dimensional

radioisotope distribution within an object. Radioisotope cameras are

mainly of the gamma ray scintillation type (Anger camera). Anger

cameras employ a sodium iodide [NaI(T1)I scintillation crystal of 1/8

to 1 inch thickness as the detecting medium. The NaI(T1) crystal is

located in back of a collimator which creates a spatial relationship

between the image and the object.










A unique new radioisotope camera that utilizes an X Y matrix of

cadmium telluride (CdTe) semiconductors as the detecting medium and

that has the detectors located within the collimating holes rather than

behind the collimator has been developed. The new camera offers im-

proved spatial resolution and reduced imaging time in comparison to the

Anger camera.

The objectives of the research described herein were as follows:

1) To develop a radioisotope camera which utilizes a small

matrix of CdTe detectors located within collimator holes.

The camera design should lend itself to use in devices

utilizing a larger matrix.

2) To describe the performance of the camera and to compare

the performance to that of other applicable cameras,

particularly the Anger camera.

3) To estimate the feasibility of larger cameras that

would be of practical clinical significance.

The design, construction and performance of the new camera is

documented in subsequent chapters. The ultimate clinical feasibility of

a larger camera incorporating the same design principles is discussed.














CHAPTER 2

HISTORICAL DEVELOPMENT


The radioisotope camera developed in this research is a unique

new design that has no direct predecessors. Some of the concepts utilized

in the camera have been tried in other types of radioisotope cameras;

however, the camera developed in this research is the first to employ

CdTe semiconductor detectors in anXY matrix and the first to locate

the detectors within the collimating holes. The research was original

in that a review of pertinent literature revealed the following:

1) Cadmium telluride detectors have not previously been

employed in an X Y matrix gamma camera radioisotope

imaging system. The idea to utilize CdTe in a radio-

isotope imaging device is not original to this research.

Work is under way at several laboratories in regard to

the development of radioisotope cameras utilizing semi-

conductor detectors (Entine, 1977). The use of CdTe

detectors in an X Y matrix, thus reducing the number of

outputs, has been offered for consideration (Allemand,

1977). One rudimentary imaging system utilizing an array

of CdTe detectors behind a parallel hole collimator was

developed (Zanio, 1977). The system utilized (n)2 outputs

(for an n x n matrix) and a nonuniform electric field

geometry. The detectors were joined electrically in an

attempt to fabricate a large volume detector. Although

the system functioned, it was not designed as a radioisotope










camera and is not really applicable to this research.

The application of CdTe to radioisotope cameras is

sufficiently diversified to assure that each study is

unique and significant.

2) Semiconductor detectors have not been employed in an inhole

matrix geometry. Cameras have been made that utilize

orthogonal strip germanium detectors (Detko, 1972). The

cameras have demonstrated significant resolution but are

somewhat limited in size. Fabrication of larger cameras from

several orthogonal strip detectors is underway. The

germanium requires cooling during operation and since each

orthogonal strip detector is actually one detector with many

leads, some cross talk between elements exists (particularly

when adjacent conducting strips are in close proximity for good

resolution). The orthogonal strip arrangement does not lend

itself to shielding between elements so as to reduce scattered

radiation contributions. The extension of the orthogonal

conducting strips beyond the ideal sensitive volume of the

detectors creates substantial nonuniform electric fields near

the edges of each detector. Within the regions of nonuniform

electric fields, the field strengths are less than in the

more sensitive regions. As a result, incomplete charge

collection occurs and substantial numbers of pulses with

reduced pulse height are produced that must be discriminated

against. The limitation on spacing between adjacent con-

ducting strips creates substantial gaps between detectors






5



that reduce the sensitive area of the matrix (in comparison

to total crystal area). Orthogonal strip cameras constructed

to date utilize large resistor divider networks to produce

one analog signal representing X position and one analog

signal representing Y position, rather than direct digital

localization of the detector in which a detection event

occurred.

3) Gamma cameras employing multiple NaI(T1) crystals have not

been substantially successful in comparison to the Anger

camera. One such device, the autofluoroscope, was composed

of 294 NaI(T1) crystals, each of which was coupled to an X

position and Y position photomultiplier tube (PMT) by long

light pipes. The autofluoroscope could operate at high count

rates but had poor photopeak energy resolution caused by

inefficient transfer of light through the light pipes (Beck,

1975). The poor energy resolution reduced the ability to

discriminate against scattered photons.

A camera, employing 254 NaI(TI) crystals, was developed

recently (Sveinsdottir, 1977). The spatial resolution was

essentially the same as previous multidetector scintillation

cameras but the temporal resolution was substantially improved

by utilizing parallel event processing in lieu of sequential

event processing. The number of crystals used remains some-

what limited by the use of one light pipe, one PMT, amplifier,

discriminator, buffer, etc. per crystal.

Another such device was the Quantascope camera which

employed 2515 cesium iodide CsI(T1) crystals that were










separated by lead strips in order to reduce the detection

of scattered radiation. The quantascope utilized an

image intensifier rather than PMTs. The camera produced

poorer contrast than is available with an Anger camera and

required matched crystal selection (Beck, 1975).

4) Another CdTe radioisotope camera was recently constructed;

however, the application involves computer-assisted recon-

struction techniques for the development of a two dimensional

representation of the three dimensional radioisotope distri-

bution (Luthmann,1978). Data is obtained from the n detector

rows situated within a parallel plate collimator which is

rotated about its central axis.

Most of the radioisotope cameras that have been developed, offer

improvement in some parameter as compared to the Anger camera. A com-

plete evaluation of the devices reveals that a multipurpose device that

can offer substantial improvement in sensitivity, spatial and temporal

resolution and hence an improvement in diagnostic information obtained

or a reduction in patient exposure, has not been developed.














CHAPTER 3

CAMERA PROPERTIES


Direct digital localization of detected events in the X Y CdTe

camera developed in this research is an advantage in comparison to the

Anger camera. Also, the utilization of semiconductor radiation detectors

(particularly CdTe) and the placement of the detectors within collimating

holes offer some advantages over conventional radioisotope cameras.

To better appreciate the advantages, some of the properties of the

Anger camera and the X Y CdTe camera will be examined.

The deposition of energy, by radiation, within the NaI(TI) crystal

of an Anger camera results in scintillations proportional to the ioniza-

tion and excitation produced. The scintillations are detected by an

array of PMTs which convert the light into electrical signals proportional

in amplitude to the quantity of light incident upon the PMTs. The

location of the energy deposition is determined via a capacitive net-

work that divides the pulse amplitudes according to the distribution of

luminescence with respect to the positions of the PMTs. The resultant

pulses are then converted to visible light flashes at appropriate sites

on a cathode ray tube (CRT). The position of the flashes on the CRT

are determined by difference circuits or ratio circuits.

For purposes of forming an image of the distribution of radio-

activity within an object, only photons created within the object that

reach the sensitive volume of the detector without scattering (coherent










and incoherent) are useful (excluding the positron camera principle).

The origin of scattered photons detected by a camera cannot be limited

to a solid angle as is possible with unscattered photons. In order to

eliminate some of the signals generated by photons that have been in-

coherently scattered (and hence have less energy), a pulse height

analyzer is incorporated in conventional cameras. The pulse height

analyzer discriminates against those signals whose amplitude is not

large enough to indicate that a full energy photon (i.e., a monoenergetic

photon whose energy has not been reduced by scatter) was absorbed by

the crystal. In practice, it is not possible to precisely discriminate

against scattered photons since even full energy photon pulse heights

exhibit statistical variation about some mean value. The signal caused

by a scattered photon or the coincident summation signal caused by

several scattered photons may have an amplitude within the statistical

variation of full energy photon signals. It is possible to discriminate

against some of the scattered photons by setting a threshold pulse

height that is just below the photopeak of the full energy photons.

Coherently scattered photons are essentially full energy photons that

are scattered into some solid angle without significant energy transfer

occurring. Coherently scattered photons do not contribute useful

information to the image formation process but cannot be discriminated

against. A collimator is incorporated in the Anger camera to create a

spatial relationship between the image created by full energy absorp-

tion events in the crystal and the distribution of radioactivity in the

object.

"Semiconductor detectors offer many advantages for clinical and

experimental uses, chiefly because of their small size and excellent










energy resolution. The energy resolution of a solid state detector

is about twenty times higher than that of a conventional NaI(T1)

scintillation detector. In addition, the elimination of scattered

radiation contributes to the improvement of image quality. The semi-

conductors are considerably more efficient than Geiger probes and their

bias voltage is lower than that required by any other probe.

At present, solid state detectors are being used in the imaging

techniques; however, thus far attempts to make a high resolution semi-

conductor gamma camera met with considerable difficulties" (Meyer, 1975,

p. 697).

Gas ionization chambers can also be fabricated into relatively

small dimensions but are limited in usefulness by the relatively inef-

ficient deposition of energy to the gas contained therein.

Attempts at utilizing semiconductor detectors in high resolution

gamma cameras have consisted thus far of placing the detector close to

the back surface of the collimator. The research described herein,

utilizes the semiconductor detectors within the holes of the collimator

device rather than in back of the collimator device.

It is important that the collimator be located as close as possible

in order to prevent photons scattered on septa surfaces from being

detected in adjacent detectors and producing counts that do not give

an accurate representation of the origin of the photon.

The in hole configuration utilizes the small size into which semi-

conductor detectors can be fabricated. NaI(TI) detectors can be

fabricated into relatively small detectors but their use is somewhat

limited by the requirement that they be hermetically sealed and the










rather inefficient transfer of their scintillations to relatively large

photomultiplier tubes via fiber optic material.

The direct electrical signal generation of semiconductors is an

important contrast to the scintillation detection process of NaI(T1)

detectors. In semiconductor detectors, the electron hole pairs created

via ionization are collected on electrodes due to the bias voltage

applied to the detector. In NaI(T1) detectors, the scintillations

produced by excitation and ionization must pass through the detector,

be transferred through the light pipe to the photoelectric material of

the photomultiplier tube where the electrical signal is developed.

Cadmium telluride is a II-VI compound semiconductor having the

sphalerite crystal structure of zincblende. Cadmium is atomic number

48. Tellurium is atomic number 52. The compound, CdTe, has an average

atomic number of 50 and a specific gravity of 5.20. The electrical

property of CdTe that makes it particularly useful as a semiconductor

radiation detector is its energy bandgap at 300 degrees kelvin (oK) as

compared to other semiconductor radiation detectors (1.5 eV for CdTe,

1.08 eV for silicon (SI), and 0.67 eV for germanium (GE)) (Palms, 1971).

The relatively large bandgap allows for operation of the detector at

room temperature without excessive thermal generation of charge carriers.

Excessive thermal generation of charge carriers results in a high

leakage current and poor signal to noise ratio which degrades detector

resolution (as occurs with Si and particularly with Ge when operated at

3000K).

The physical properties of CdTe that make it particularly useful

are its relatively high atomic number (Z) (CdTe:50, Si:14, Ge:32) and










its relatively high specific gravity (CdTe:5.20, Si:2.4, Ge:5.35). The

relatively high Z and specific gravity enhance the energy absorption

efficiency of CdTe.

The CdTe detectors used in this research are p type, chlorine doped

for compensation and have chemically deposited platinum electrodes.

Platinum is frequently used as the electrode material since its work

function is relatively high which helps to prevent polarization phenomenon

(Siffert, 1976).

The electron and hole mobilities are on the order of 1000 and

80 cm2/V-sec, respectively. The electron and hole mobility-trapping
-3 -4 2
time products are on the order of 10 and 10 cm /V respectively. The

detectors are characterized as homogeneous semiconductor radiation

detectors. The crystallographic orientation of the crystals to be used

is not precisely known. The crystals grow preferentially in the [111]

direction. Detector slices are cut perpendicularly to the direction of

growth and electrodes are applied across what is probably the [111]

principal axis. "The effect of crystallographic orientation of the

detector was examined for the applied electric bias along three

principal directions--[lll, [110] and [100]--but no significant dif-

ferences were observed" (Siffert, 1976, p. 159).

The pulse shape produced by the detectors is a function of the

position and orientation of charges liberated. Ideally the charge

liberation would be parallel to the electrodes and very near the negative

electrode, resulting in rapid, simultaneous collection of all charges.

If the charge liberation is perpendicular to the electrodes, the charges

nearest the electrodes are collected first and so forth, such that










charge collection occurs over some time period. For charge liberation

near the negative electrode, high mobility elecrons traverse the

entire detector thickness and the low mobility holes traverse a much

shorter distance; hence charge collection requires less time than

those events in which holes must traverse longer distances. Ideally

the CdTe detectors should be irradiated with a normal incidence on the

negative electrode.

The approximate physical dimensions of the detectors available for

this research were 2 x 2 x 7 mm. The electrodes were two opposed sides

of 2 x 7 mm dimensions. The linear attenuation coefficient of CdTe

is approximately 0.36/mm for 140 keV photons and hence a 2 mm thick

detector should attenuate 60 percent of incident 140 keV photons (Hoffer,

1971). Photons of 140 keV are considered throughout this dissertation

since they are produced in the decay of technetium-99m (Tc-99m) which

is widely used in radioisotope imaging studies. The 7 mm detectors

will be cut into 3 detectors of approximate dimensions 2 x 2 x 2 mm.

The 60 percent attenuation of the photons of interest is sufficient

to demonstrate the characteristics of the X Y CdTe camera.

The stability of the detector characteristics with time is important

since polarization phenomenon (i.e., changes in the internal electric

field with time) have frequently been found in CdTe detectors. It has

been reported that some p type, chlorine doped CdTe detectors with

platinum contacts are totally stable with time, do not exhibit polariza-

tion and are sensitive throughout their volume (Serreze, 1974).

The bias voltage applied to the detectors will determine the

mean free path of the electrons and holes. To maximize charge collection,










the mean free path of the charge carriers should be less than the col-

lection distance or detector width. For the best photopeak energy

resolution in a 2 mm detector thickness, the bias voltage applied should

be on the order of 500 volts. The detector generated noise due to

increased leakage current caused by electric fields as high as 2500 V/cm

make operation impractical at that level. It is recommended that the

bias voltage be maintained at the maximum practical voltage (Malm,

1973). For a practical bias voltage of only 500 V/cm, the charge col-

lection time for a 2 mm detector would range from approximately 0.4 psec

to 5 isec. The decay time of a NaI(T1) scintillator is approximately

0.3 psec.

The detector thickness is essentially limited by the field strength.

Reasonable spectrometer performance for the CdTe available occurs

around 325 to 500 V/cm. For a minimum trapping time of 6 psec and

mobilities of 80 and 1000 cm2/V-secfor holes and electrons respectively,

the detector thickness is limited to about 2 mm. For higher field

strengths, noise degrades the spectrometer performance. Further

development of CdTe could enable higher spectrometer field strengths

and hence thicker detectors.

Cadmium telluride has many other properties which are not of great

interest in semiconductor radiation detector applications; however,

for completeness some of those properties are listed as follows:

piezoelectric, photovoltaic, magnetoresistive, pyroelectric, electro-

optic (Aven, 1967). A listing of the main applications of CdTe is as

follows: infrared windows, electro-optic modulator, Gunn effect high

frequency current oscillator, piezoelectric devices, luminescent devices,

solar cells and nuclear detectors (Wald, 1977).










Other high density compound semiconductors are currently avail-

able and the expectations for other high Z semiconductor radiation

detectors have been evaluated (Armantrout, 1977). Aluminum antimonide,

indium phosphate, and zinc telluride have had relatively little work

accomplished and ultimate technological feasibility is yet to be deter-

mined. Mercuric iodide and gallium arsenide have been fabricated and

used, but do not excell in comparison to CdTe.

The electrical signal output of the X Y matrix is developed as

illustrated in figure 1 for a 25 element matrix. Electrical signals

are developed directly via the semiconductor radiation detector principle

rather than the less direct electrical signal generation of scintillation

radiation detectors. The X Y matrix signal configuration reduces the

number of outputs for an n x n matrix to a minimum of 2 n. Outputs

numbering in excess of 2 n may be required since the summation of

capacities or leakage currents in any one column or row may limit the

number of detectors per output.

The theoretical limit of resolution for the camera approaches the

centerline to centerline distance between adjacent matrix elements. The

ultimate resolution of a camera of this type can be selected to the

extent that the size of the detector and the size of the collimating

holes can be selected. Of course, as the physical size of the matrix

elements is reduced, the number of matrix elements and the associated

electronics are increased proportionately.

The sensitivity of the camera ultimately depends upon the duration

of pulses produced by the detectors and associated electronics and upon

the number of matrix elements. Large cameras would be composed of







YAW-) U5N
Ak A, A,


Fim 1 CoTe Mucum Muax


"f










several matrices where each matrix has its own electronics package.

Electronic storage devices are available that can temporarily store the

data as fast as they are produced. The size of a matrix is essentially

governed by the number of detectors that can be operated

without serious degradation of signal to noise ratio. Improvements

in detector properties would produce analogous increases in matrix size

possible.


Advantages of a CdTe Matrix


The application of CdTe as semiconductor radiation detectors in an

X Y matrix radioisotope camera eliminates some problems of systems in-

corporating NaI(T1) detectors. The advantages of CdTe in an X Y matrix

detector are summarized as follows:

1) "The elaborate assignment of X and Y coordinates in the

scintillation photomultiplier array method is completely

eliminated by the simple but ingeneous method of reading

these coordinates directly off the conduction surfaces of

the semiconductor itself, using the direct, spatially

quantitized digital output of the crystal" (Beck, 1975, p. 25).

The development of X Y positional information in the

Anger camera utilized analog signals to locate the center

of luminous intensity. As described in 2) below, the method

creates some inherent limitations in regards to determining

the actual position of radiation interaction in the NaI(T1)

crystal. An X Y matrix of CdTe detectors can be used to

obtain a digital address (row number and column number where

signals were generated by a detection event) and the










amplitude of the signals produced can be analyzed by

pulse height discrimination to eliminate some of the events

caused by scattered photons.

2) The X Y CdTe camera will not produce distortion near the

edges of the matrix. The generation of X Y positional informa-

tion by the capacitive network pulse divider of the Anger

camera is based on the center of luminous intensity incident

on the PMT array. For detections that result in sufficient

luminescence production to be indicative of full energy

photons, events are displayed on the CRT in positional rela-

tion to the effective center of luminescence incident on the

PMTs. Due to inhomogeneities and defects in the NaI(T1)

crystal and due to reflection of light at the crystal

boundaries, the center of luminescence incident on the PMTs

is not in the exact position of energy deposition by the

radiation. This distortion is relatively small in the

center of the crystal but increases near crystal edges. The

distortion is as high as 30% near the edges of some Anger

cameras (Moretti, 1974). Photons that undergo multiple

incoherent scatters within the NaI(TI) crystal can cause

significant differences in the location of the center of

luminous intensity and the point of the photon's initial

interaction (which contains the positional information).

For a matrix of CdTe detectors, the location of a

photon detection is determined to be within a particular

detector via the X and Y address produced. The positional

information will not be distorted near the edge of the matrix










since the output is an address developed by the detector in

which the energy is deposited. The detectors are separated

from each other by the collimator septa and hence the probability

that a photon can deposit energy in more than one detector

is substantially reduced.

3) Improved photopeak resolution is possible with CdTe. In

scintillation detectors, the deposition of energy in the crystal,

by radiation, results in ionization and excitation. The energy

is converted to light by the luminescence process. The light

travels through the crystal and by means of a light pipe and

reflectors is transmitted to the photocathode of the PMT. The

light is absorbed by the photocathode resulting in photo-

electron emission. An electron multiplication process

produces a large pulse at the output of the PMT. Each step

in the process has its own efficiency and statistical varia-

tion. The statistical variation of the PMT output pulse is

the propagation of the variations of the individual steps

involved.

The semiconductor radiation detection principle of CdTe

is somewhat simpler in that the ion pairs (created by energy

deposition) are collected at the electrodes due to the electric

field developed across the detector by the bias voltage

(assuming trapping and recombination are prevented). The con-

version of ion pairs to light and light to electrons and the

associated statistical variations are eliminated. The charge

collected at the electrodes is amplified into a large pulse.









The advantage of CdTe is demonstrated by a comparison

of photopeak energy resolutions. The full width at half

maximum (FWHM) energy resolution for a 122 keV photopeak on

Anger cameras is on the order of 14 to 20 keV (Allemand, 1977).

The FWHM energy resolution of CdTe for 122 keV photons has

been reported to be as low as 3.8 keV (Siffert, 1975). For

the CdTe available for this research, the FWHM is on the order

of 8 to 10 keV for 140 keV photons.

4) In addition to the potential for improvement in pulse height

resolution described above, CdTe has a higher density and

higher atomic numbers resulting in increased photon attenua-

tion per unit of detector thickness.


Advantages of In Hole Detectors


The advantages of locating the detectors within the collimator holes

rather than behind the collimator are as follows:

1) As discussed previously, discrimination against scattered

photons is desired. The discrimination is somewhat complicated

since full energy photon pulse heights vary statistically

about some mean value. Ideally, pulse height discrimination

would eliminate from consideration all scattered events due

to the reduced energy of the photons; however, there is a

statistical fluctuation around the true pulse heights that

complicates total discrimination against scattered photons.

For individual CdTe detectors the full width at half maximum

is on the order of 8 10 keV for 140 keV photons for the

detectors used in this research. When several detectors are










operated in a row, the errors associated with noise are

propagated and the FWHM is increased. For existing Anger

cameras the FWHM is on the order of 14 to 20 keV for 140 keV

photons. Since the average energy loss in one incoherent

scatter by a 140 keV photon is only 24 keV and occurs at a

photon scatter angle of approximately 750, there exists a

significant fraction of photon scatters that result in

relatively small energy loss with significant scatter angles.

The most probable photon scatters depart very little energy

and have small scatter angles. Scatter angles of 900 are

the least probable (approximately 33% of maximum probability

and have scattered photon energies near 110 keV). As the

scatter angle increases from 900 to 1800, the scatter probability

associated with the angle increases from approximately 33% to

approximately 55% of the maximum probability.

For photon scatter events that occur in one detector,

adjacent detectors occupy 20% to 50% of the solid angle that a

scattered photon might enter. Since pulse height discrimination

cannot totally delete contributions from scattered photons,

individual CdTe detectors can be placed within the holes of

the collimator and hence the detectors can be separated by

collimator septa. The septa can then significantly attenuate

scatter occurring on the collimator hole surfaces or within

the detectors themselves, that might otherwise be counted in

adjacent detectors and contribute spurious data to the image.

Discrimination against pulses that have amplitude in excess

of the photopeak pulse height resulting from the coincident










summation of two or more detection events are not necessary

in this camera since detection rates will be maintained at a

low level.

The advantage is further illustrated by an examination

of applicable attenuation coefficients. For 150 keV photons,

the attenuation in tin due to scattering (including coherent)

is approximately 25% of the total attenuation coefficient

(Hubbell, 1969). Values for tin are used, since the average

atomic number of CdTe and tin are the same. One of four

photons interacting in CdTe will undergo scattering. If the

CdTe detectors are not located in hole, many of the scattered

photons could be detected in adjacent detectors and hence

produce erroneous positional information. The reduction of

radiation scatter between detectors by the in hole detector

principle, is particularly important in this application since

the small size of the detector and the size and cost limitations

on amplifiers may make precise pulse height discrimination

against scattered radiation very difficult.

The fabrication of the camera is not greatly affected by

the in hole detector geometry since the detectors must have

electrical leads applied, must be separated by insulating media

and have regular overall physical dimensions for ease of

construction. It has been demonstrated that separation of

multiple NaI(T1) detectors by absorbing septa reduces scatter

and produces finer resolution (Beck, 1975).

2) As shown in figure 2, for a point source located on the

collimator hole axis, the detection of primary or full energy











MAUDERLI Walter
429-72-4216


DETECTOR-COLLIMATOR ARRANGEMENT
I FOR


I I


2 POINT SOL
I I


JRCES

ANGER TYPE
CAMERA

LEAD HOLES

-- ,-DETECTOR


S: Scatter orig. in" Overlapping sector of primary photon beam
detector reducing resolution
R: Scatter orig. in -reducing resolution*
collimator

1. \ PARALLEL
TU N i PLATE TYPE
TUNGSTEN \ I RCAM E
PLATES I CAMERA
E DETECTOR


None of the I
above Problems


i


* Pulse-height discrimination does not eliminate scatter radiation when
the scatter angle is less than the following values: In case of Tc-99m
Lower discriminatory level 5% below photon energy: I 50 (tor Compton
10% N 70 scatter)
20% : 100 WM 012578


DETECTOR-COLLIMATOR ARRANGEMENTS


.. .


j~f


I


FIGURE 2










photons is essentially restricted to within the physical

dimensions of the detector located in the collimator hole.

For a NaI(T1) crystal located behind the collimator, a detec-

tion could occur beneath an adjacent collimator hole.

Scattered photons (scattered on collimator surfaces or within

the detector) are significantly attenuated by the collimator

septa that separate individual detectors.


Disadvantages


The application is not without some limitations. The disadvantages

are as follows:

1) A substantial gap will exist between adjacent detectors

resulting in a reduced detector area as compared to a solid

crystal such as an Anger camera employs. The gap produces a

sensitive detector area of approximately 50% of the overall

camera dimensions as compared to 65 to 80% for the Anger camera.

The sensitivity is therefore reduced somewhat (for equivalent

detector thicknesses). The ratio of the sensitive detector

area to the camera face area could be increased with more

sophisticated fabrication technology.

2) The theoretical limit of resolution is limited by centerline

to centerline spacing of the detectors. In this application,

the centerline to centerline distance may be in excess of 3 mm.

Should the application prove successful, the distance could be

reduced by smaller detectors and gaps.













CHAPTER 4

CAMERA DESIGN AND CONSTRUCTION


An X Y matrix radioisotope camera was designed and constructed in

accordance with objectives set out in previous chapters. A block

diagram of the resultant camera is presented in figure 3. In the fol-

lowing sections, each component of the camera will be discussed in

some detail, with emphasis upon the objectives for the design and con-

struction of each component.

Detector Matrix


The detectors utilized in the matrix were cubes of approximately

2 mm side length. The detectors were cut from large crystals

(2 mm x 2 mm x 7 mm) using a wire saw (0.003 diameter stainless steel

wire impregnated with 8 micron size diamonds). The crystal surfaces

were not treated (lapped or polished) after cutting (some of the

large crystal surfaces were not treated either). The cubes had

platinum electrodes applied to an area on two parallel opposed sides

(the electrodes did not extend to the edges of the crystals on the

large crystals or on the small cubes). The large detectors that were

cut to produce the small cubes were selected for the lowest noise,

best photon spectra and best physical characteristics (i.e., a

regular shape with electrode surfaces approaching crystal edges)

from approximately 130 available.










































Fie.. 3 hm DiCmm OF X Y CaTE CEmm




Ul










Substantial work was done in developing cutting techniques to

minimize degradation of spectrometer performance during cutting. Noise,

photon energy spectra and stability with time were examined for each

large crystal prior to cutting and for the small cubes after cutting.

Figure 4 represents a typical energy spectrum for a large crystal prior

to cutting and one of the small cubes after cutting (the noise voltage

and leakage current are also represented).

Wire leads were attached to the platinum electrodes with conducting

silver epoxy. Attempts were first made at mechanically coupling a wire

spring to the electrodes, but the hardware required could not be easily

manufactured. Attempts to use a silicon rubber based conducting

compound were also futile since the resultant bond between the wire and

the platinum was not adequate to allow handling of the crystal by its

wire leads. Ultrasonic bonding was also investigated but disregarded

since it requires high local temperatures and does not work well on

platinum surfaces. The wire used in ultrasonic bonding would not be

strong enough to allow for handling the crystal by its wire leads.

The epoxy utilized is a two component silver filled epoxy designed

specifically for chip bonding in microelectronic and optoelectronic

applications. It cures at temperatures as low as 50C (for 12 hours).

A detector, prior to wire lead bonding, is shown in figure 5 and after

wire lead bonding in figure 6.

A printed circuit board was designed that would facilitate the

coupling of rows to form X signals and the coupling of columns to form

Y signals (see figure 7). The board is a double sided printed circuit

board incorporating plated through holes. The wire leads of the detectors

were soldered to the printed circuit board. A layer of teflon was













DETECTORS: CDTE
ISOTOPE: Co-57
BIAS: 100


PREAMPLIFIER: ORTEC 112A
AMPLIFIER: ORTEC 450(1.5SSEC RCs)


ORIGINAL CRYSTAL PRIOR
To CurrTTr(ND035-6-3)

LEAKAGE CURRENT: 25mAa65V


E EMYWIY


ONE OF SMALL CRYSTALS
AFTER CUTTirr (ND035-6-3-A)
LEAKAGE CURRENT: 6lNA65V


SIr KEY


FIGURE 4 SPECTOETRY PERFORMANCE BEFORE AND AFTER CUTTING


27





































FIGURE 5


CDTE DETECTOR (2x2x2 MM APPROXIMATELY)




































FIGURE 6


CDTE DETECTOR (WITH LEADS APPLIED)




















ToP


FIGURE 7


BOTTOM


PRINTED CIRCUIT BOARD


U










inserted between the detectors and the board for electrical insulation.

The printed circuit board has an edge connector that facilitates the

application of bias voltage and access to X and Y signals.

For convenience, the electrodes were oriented parallel to the

surfaces of the collimating holes. Ideally, the electrodes should be

oriented perpendicular to the collimating holes so that the negative

electrode is irradiated (as explained in Chapter 2). The operating

characteristics of each detector were tested for both possible bias

voltage polarities. Each detector was ultimately operated in its

optimum bias voltage polarity.


Collimator


The research involved the design and fabrication of a multihole

collimator. A multihole collimator was selected since it offers higher

sensitivity for a given resolution than the pinhole collimator, is useful

for large organ studies and is not subject to fall offin perimeter

response as is the pinhole collimator (Anger, 1975). A parallel hole

collimator was selected since the size of the image is essentially

independent of the distance from the subject to the collimator and for

a given resolution at a specified depth has better depth of focus (Anger,

1967). The most commonly used collimator is the parallel multihole

collimator. Although parallel multihole collimators were selected for

this research, the application of CdTe or other semiconductor radiation

detectors could be extended to focused multihole collimators and even

to pinhole collimators. For focused collimators, the fabrication is

more difficult, but the in hole detector approach could still be utilized.

For pinhole collimators, the in hole detector approach could be applied










if the hole length were equal to the detector length so that scatter

between detectors would still be essentially eliminated.

The collimator holes are square since the cross section of the

available detectors is roughly square. CdTe cannot readily be made

into small diameter rods without inducing strains (ESPI, 1969). The

square collimator holes facilitated design calculations since septal

thickness is constant. The material selection for the collimator is

based on atomic number, density, machinability and expense. Tantalum

was used as the material due to its high Z (73), high specific gravity

(17.1), ability to be drawn into square tubes of appropriate dimensions

and relatively low expense. It is nontoxic. The collimator and

detector configuration were developed for use with the 140 keV photons

of 99mTc.

The square hole size was maintained at the minimum consistent with

the insertion of the detector and associated electrical leads and insula-

tion. The detector electrodes were situated parallel to the hole axis.

If the electrodes had been perpendicular to the hole axis, the capacity

between the collimator and the detector electrodes would have been

minimized. If the electrodes are parallel to the hole axis, the

detector length and hence to some extent the sensitivity, could be

varied; although the capacitance may be increased and resolution de-

creased.

Design criteria for parallel multihole collimators as used in con-

junction with an Anger camera are available (Anger, 1967; Causer, 1975).

The design criteria assume that the effective center of the detector is

located at some distance away from the back of the collimator. The










effective centers of the in hole detectors are located within the

collimator and hence existing design criteria are not exactly analagous.

Theoretical parallel multihole collimator designs attempt to optimize

the number of holes, their size, shape, septa thickness, and hole length.

Practical collimator design parameters have largely been assigned

empirically (Myhill, 1967). Since the scope of this research was

primarily involved with the validity of the application rather than the

optimization of the collimator dimensions, selection of collimator dimen-

sions was based on existing criteria for Anger type parallel hole col-

limator design. Causer (1975) indicates that optimum collimator length

is essentially independent of resolution and that for 99mTc photons,

optimum collimator length is approximately 1.8 2.5 cm. Collimator

length was selected as one inch since the effective collimator length

is reduced somewhat when the detector is located in hole. For a square

hole side dimension of about 0.115 inches, according to Anger (1967),

the septum thickness should be approximately 0.0081 inches. The actual

collimator dimensions are: length--l.0 inches (.01 inches); septa

thickness--0.008 inches ( 10%); square hole side dimension--0.115

inches (+ 0.001 inches).

The shortest distance a photon can travel through a septum when

taking the unwanted path of minimum attenuation through the collimator

is approximately 0.0355 inches. For 99mTc photons in tantalum this is

equivalent to an attenuation of 97%. The attenuation of scatter between

detectors for scattered photon energy of 140 keV is approximately 50%

and for scattered photon energy of 100 keV is approximately 80%.

The square tantalum tubes are assembled inside of an aluminum block

(see figure 8). Most of the tantalum tubes were larger than the square






34


FIGURE 8 COLLIMATOR










dimension tolerance and hence it was necessary to hand select tubes

and heat press them into the block. The component parts of the detecting

head (less detectors) are as shown in figure 9. A sleeve of insulating

tubing was placed around each detector when inserted into the collimator

so as to electrically insulate the detector from the tantalum tubes.

An aluminum foil was placed over the top of the collimator so as to

maintain the detectors in a dark environment (operation in light fields

increases the detector leakage currents). Additional layers of lead were

added to the exterior of the collimator to provide additional shielding.

Analog Circuitry


The analog portion of the electronics (i.e., that portion that

collects the charge from the rows and columns connecting the top electrodes

and bottom electrodes, respectively, and converts the charge to a voltage

representative of the energy deposited in a detection event) was

designed to meet the following criteria:

1) To produce a voltage pulse that is proportional in amplitude

to the charge pulse at the amplifier input and hence to the

energy deposited in a detection event.

2) To minimize the number and cost of components so as to

facilitate the hybridization that would be required for

clinically significant devices.

It is desirable to minimize the number of components since an n x n

matrix would require a minimum of 2n amplifiers. A clinical device

would require a large number of amplifiers and therefore in order to

minimize cost and to facilitate design hybridization, it is desirable to

use the fewest components commensurate with the application.






























A-BASE PLATE
B-SPACER BLOCK
C-TEFLON INSULATOR
D-PRINTED CIRCUIT BOARD


FIGURE 9


E-TEFLON INSULATOR
F-COLLIMATOR
G-COVER PLATE


DETECTOR HEAD (LESS DETECTORS)










In keeping with these ideals, a one stage amplifier that has

sufficient amplificationto convert a relatively small charge pulse

collected from the crystals (on the order of less than 0.005 picocoulombs

for 100 percent charge collection from a 140 keV photon) to a voltage

pulse that has sufficient magnitude to be used directly in an analog to

digital conversion device was developed (see figure 10). It has been

reported that the actual charge collection efficiency in CdTe is sub-

stantially less than 100% and is dependent upon electric field strength

and energy deposition location (Malm, 1973). The amplifier is essentially

a charge sensitive amplifier or charge to voltage converter. A primary

advantage of the charge sensitive design is that the output is independent

of detector capacity (which may vary with time and number of detectors

per row). The detector is alternating current (a.c.) coupled to the

preamplifier in order to remove the direct current (d.c.) component prior

to amplification. Direct current coupling would be preferred since it

is somewhat less noisy, but in this application, signals are acquired

from both of the detector electrodes which requires that at least one

side be a.c. coupled in order to remove the bias voltage component

applied to the detector. The amplifier consists of an operational

amplifier with a junction field effect transistor (JFET) at the input.

The resultant voltage output (V) is related to the charge input (Q) by

the expression:
Q
V= C

where Cf represents the capacity between the input and output of the

amplifier. In order to maximize the voltage output for a given input,

the capacity Cf should be minimized, thus no added capacity was utilized




























































-M


Ftis 10 Aim. CIacuITR


38


GQwen
#A nsal










between the input and output and the feedback capacity is the stray

capacitance between the input and output. A feedback resistor is

required to discharge the feedback capacity; otherwise, the feedback

capacity would build up a charge when input signals are present that

could eventually saturate the amplifier. A high value feedback resistor

is utilized in order to maintain a large time constant to maximize

the collection of charge deposited within the crystal. For the CdTe

detectors utilized in this research, an electric field of about 325 V/cm

gave optimum spectrometer performance. Since the electron and hole

mobilities of CdTe are on the order of 1000 cm2/Vsec and 80 cm2/Vsec

respectively, the charge collection time for a 325 V/cm field strength

is approximately 0.6 psec for electrons and 8.0 usec for holes. In

order to assure that the collection of positive charges (holes) and

negative charges (electrons) is maximized (in order to maximize the

amplifier output voltage per detection event), a 500 Ml resistor was

utilized in the feedback network. The large feedback resistor thus assures

that the amplifier time constant is much much larger than the transit

time for hole collection and hence maximizes the charge collected per

detection event. The operational amplifier was selected for its high

gain and relatively low noise. The signals to be amplified occur at a

relatively low frequency (due to relatively long charge collection times

and due to a relatively low contemplated patient flux incident upon the

matrix), and hence the frequency--bandwidth characteristics of the

operational amplifier are not of primary importance. The JFET employed

at the input of the amplifier, determines the amplifier input impedance

and develops the signal/noise ratio that will be evident at the output

of the amplifier. The operational amplifier primarily provides the gain.










The placement of an external JFET at the input alleviates somewhat

the requirements for high input impedance for the amplifier. The

amplifier selected has relatively good specifications for all of these

characteristics but excels in gain.

Although many operational amplifiers have field effect transistors

at the input, an external junction field effect transistor was utilized

in this amplifier since they exhibit much lower noise figures than

insulated gate field effect transistors (metal-oxide-silicon-field-

effect-transistor). The use of a discrete FET yields improved noise

characteristics in comparison to the fabrication in a monolithic design.

The particular JFET selected has a high common source forward

transconductance to common source input capacities ratio (Elad, 1969).

The ratio determines essentially the magnitude of change at the output

of the JFET per change at the JFET input. The JFET also has low noise.

A passive capacitor resistor network was added at the amplifier output

to differentiate the pulse to produce a pulse duration on the order of

50 psec or less. The resultant pulse is bipolar but this is of little

consequence in this application since the output pulse is fed to the

analog to digital conversion device which is sensitive only to the

magnitude of the leading edge component of the pulse. The capacitor

resistor network also serves to remove the direct current voltage level

present at the output of the amplifier. A direct current voltage level

present at the output would complicate the design of the analog to digital

converter.

The power inputs to the amplifier are filtered by a resistor

capacitor network that serves to bypass ground loops and to smooth (by

integration) voltage fluctuations caused by wideband noise pickup.










This research utilized a 5 x 5 matrix of CdTe detectors and hence

required 10 amplifiers. The 10 amplifiers were fabricated onto one

printed circuit board and hence it was necessary to pay particular

attention to guarding the inputs of the amplifiers to minimize feedback

caused by voltage and current fluctuations in adjacent amplifiers.

A state of the art commercial preamplifier and amplifier were procured

for use in testing the CdTe detectors. The commercial amplifiers offer

low noise and large signal to noise ratio and the ability to control the

integration and differentiation of the signal. When using the commercial

amplifiers, the photopeak energy resolution is largely determined by the

noise generated within the detector. The commercial amplifiers were

used as a reference device for testing the detector properties and for

comparison of the inexpensive amplifiers developed for use with the

matrix.

As shown in figure 11 and figure 12, the low cost charge sensitive

amplifiers developed in this research offer signal to noise ratio and

photopeak resolution comparable to the state of the art commercial

amplifier.


Analog to Digital Circuitry


The analog to digital converter was designed to produce a digital

output signal whenever an input signal exceeds some reference level and

hence indicate that a detection event occurred. The reference level can

be adjusted to represent the lower limit or threshold of a photopeak and

hence pulse height discrimination can be performed. The digital outputs

of the analog to digital converter circuits are inputs to the digital



















M AUDERLI Walter -
429-72-4216

COMPARISON OF OUR SPECIALLY DESIGNED
PRE-AMPLIFIER WITH A COMMERCIALLY
AVAILABLE INSTRUMENT

Detector: CdTe
Isotope : Co-57
ORTEC Pre-Amp.
Mod. 142-A
(followed byOrtec
Research Amp.
2 Mod.450)

Pre-Amp. Cost:
m 3 495.00
a.s

S/N v 6.5

Time N



SPECIALLY DE-
SIGNED Pre-Amp.
(followed by Ortec
Research Amp
2. Mod. 450)

Cost of Material:
S 5 25.00
l L S/N 6.3

N
Time

cont.onnext page w Ior2o78


AMPLIFIER SIGNAL-TO-NOISE COMPARISONS


FIGURE 11







43







N1At'DERLI Woalter
42i- 72 -4216

Detector : CdTe
Isotope: Co-57






ORTEC Pre-Amp.
Mod. 142-A
(followed by Ortec
z Research Amp
Mod. 450 and
STracor Northern
Multichannel Analyzer
Mod. TN- 1705
1024 Ch.)
o
0



ENERGY
122 136 keV



SPECIALLY DE-
SIGNED Pre-Amp.
(followed by the same
z Instrumentsas above)
z
I
0

z





ENERGY
1E2 136 keV


WM 012078


AMPLIFIER SPECTRUM COMPARISONS


FIGURE 12










logic circuitry. The converter was also designed to facilitate hybridiza-

tion. Since the output of amplifiers servicing rows is of opposite

polarity to the output of amplifiers servicing columns, it was necessary

to develop converters for negative input signals and positive input

signals, that would produce a positive logic pulse when the input signal

exceeded some reference level.

The converters consist of voltage comparators as shown in figure 13.

The output of the comparators is fedback to the input in order to create

hysteresis and hence prevent input signal noise from causing the output

to switch states several times as the input signal passes through the

reference voltage level. Hysteresis serves to speed up the circuit

switching time and cause the comparator to switch only once as the input

signal passes through the reference voltage level and hence prevent

chatter. Chatter is deleterious to this application since it could

ultimately be evaluated as multiple detection events. The voltage

comparator was selected because it has a fast response time and reasonable

input offset and input bias currents. Capacitors were added from the

outputs of the comparators to ground in order to alleviate oscillations

caused by the output signal from one comparator being fedback by stray

coupling to the input of an adjacent comparator. The comparators were

assembled onto a printed circuit board but the inputs were inadequately

guarded to prevent the oscillations. The comparator printed circuit

board was shielded from the amplifier printed circuit board (and other

boards) in order to alleviate oscillation problems encountered due to

feedback caused by stray coupling.



























Digital
LM319 Output



125pF



S-15V

For Positive Analog Input Signals
(Voltage supplies are filtered as shown on figure 10.)




+15V







(+ Digital
LM319 Output



< == 125pF



150k








For Negative Analog Input Signals
(Voltage supplies are filtered as shown on figure 10.)


ANALOG To DIGITAL CONVERTERS


Analog
Input


FIGURE 13









Digital Circuitry


The digital circuitry was developed to meet the following criteria:

1) To identify, based on the digital outputs of the row

and column circuitry the individual detector in which

a detection event occurs and to make that data

available for input to a microcomputer.

2) To eliminate from consideration, those events in which

more than one row signal or more than one column signal,

or both, occur within some coincident time frame.

The digital circuitry was designed such that the only events that

are collected as data are those in which one row signal and one column

signal occurred simultaneously (within some coincident time period) and

hence identify that a detection event occurred in a particular detector

(see figure 14). When a data input line goes high (after some quiescent

period when all inputs are low), the circuitry (after a brief wait for

coincidence), latches the data present. If the data consist of one X

signal and one Y signal, then that data are entered into a storage

device for subsequent transfer to a microcomputer port. If the data

are not acceptable or after completion of the data transfer, the process

starts again. The time frame in this particular device is not of

great importance since there are only 25 detectors supplying signals

and since the microcomputer requires a relatively long period of time

to input data.

The dual monostable multivibrators (one shots) A, B, C, D and E

serve to produce one 9.1 usec pulse each time that any of the X1 X5

comparators or Y1 Y5comparators produce a transition from logic 0 to

logic 1 at the one shot input while the one shot output is low. The





















































A.IBC.*.3*F.T.i D 4mr
I,N,c. , a 407
L,1,1.W Qa
IJA. U o 404

o,,Q. a M1L


n a usa
.1 a uM



7w,0 m


&"1 4 IMpt M eLL
Qta ele a laet
Ia bedba



a4 l e-4 Abt fm flet
M t** utl nme a t, n
snl ~mmer )
tees a awe eas"
Q^J | fMY 1MBI
Nrt


FIGURE 14 DIGITAL CIRCUITRY (PAGE 1 OF 2)





























































ruam
Del a dot
Wal 4 i*pt m r
Qud cloehrd a latch





sal 4 Imput 8ate
64 -w d bit firt la first
ot serial nmiory(FTN)
Q4d 2 -0p iM late
No~de


FIGURE 14


DIGITAL CIRCUITRY (PAGE 2 OF 2)


C.I.C. T.

IJ,I,JFFK.O
L.N.AA.M
N.T
,P,.Q,I
I
6.1
cc,=


7W_


OML
CD 40
a 4072
CO 4042
C- 6050
C wn
CD 4061
C, 4011
C. 4462
3341

13as o










9.1 psec period of these one shots allow sufficient time for subsequent

data processing to be performed prior to the consideration of another

event. The one shots do not reset upon the application of each new

pulse and hence additional input pulses applied to a one shot during

the high period of the one shot will be ignored. The one shot outputs

are fed to OR gates G and H and to latches I, J, and K. When pin 1 of

OR gate G goes from low to high, one shot F enables (after a coincident

time delay of up to 5 Vsec) the latches I, J, and K. The data present

at the latch inputs are held at the latch outputs for the 2.0 usec of

the enable pulse (which allows time for subsequent data processing). It

is important that the only time that one shot F will trigger is after

one shot A, B, C, D and E outputs have all been low and then one or

more low to high transitions start the analysis of another event.

Once the data have been latched in I, J, and K, they are buffered

through buffers L and M.

The output of the buffers is fed to AND gates 0, P, Q, R, OR gates

U and V, and NAND gate S, which in combination with NOR gates N and T

and AND gates W and X produce a high output at pin 1 of AND gate W if

data are present and the data consist of one X and one Y signal.

The outputs of buffers L and M are also fed to a diode matrix that

reduces the X data from 5 lines of decimal data to 3 lines of binary

data and likewise for the Y data. The diode matrix design could be

directly upgraded to a 16 x 16 matrix, by reducing 16 decimal lines to

4 binary lines on the X and Y axes and utilizing all 8 of the parallel

port input lines (rather than the 6 used in this application). When

pin 1 of AND gate W goes high, one shot Y is triggered, which enables

latches FF and GG to latch the data present on the X and Y binary data










lines. The output latches of FF and GG are buffered through buffer BB

to the inputs of first in first out serial memories (FIFO) CC and DD.

The FIFOs serve to assemble incoming data at one clock rate and transfer

the data out at a different clock rate. Each FIFO is a 64 word x 4 bit

storage device. Attempts to input data into the FIFO when all 64

registers are full are ignored. The FIFOs can easily be ganged to create

a composite FIFO of any dimension desired. When high, pin 1 of AND gate

W (after a delay through one shot AA for data set up) enables FIFCs CC

and DD to latch the data at their inputs. After data have been accepted

into the serial memory of the FIFOs, the FIFOs generate through NAND

gate EE a reset pulse that resets one shot Y to be ready for the next

data transfer.

Once data are entered in the FIFOs, data automatically "bubbles"

to the FIFO output. The microcomputer parallel input port (an eight

bit port) has two control signals that readily interface with the output

control lines of the FIFOs. The output ready lines of FIFOs CC and DD

activate NAND gate EE to create through buffer AA and one shot HH, a

lu sec pulse that pulls low the parallel data ready (PDR) line of the

microcomputer indicating that valid data are present at the port. The

microcomputer, under program control, inputs the data from the parallel

port and stores the data in memory for subsequent processing. Once

the data transfer into the microcomputer is complete, the parallel

input acknowledge line (PIAK) goes low to signal the FIFOs that the

data transfer is complete and that another data transfer may be initiated.

A master reset signal is activated by a switch, through one shot Y.

The FIFO memories are cleared and the FIFO control lines are initialized

when the master reset signal is activated.










The time frame for selection of a valid event is not very critical

in this application since the microcomputer uses in excess of 33 Psec

per data transfer. Complimentary metal oxide semiconductor (CMOS) devices

were utilized since they have low power consumption and good noise im-

munity and since the data transfer time is not critical. Processing

time in the digital circuitry could be greatly reduced by employing

transistor-transistor logic (TTL) devices throughout and by incorporating

faster FIFOs.


Microcomputer

The microcomputer was selected to meet the following criteria:

1) To have sufficient hardware capabilities to readily input

data from a parallel port and store the data in memory with

minimal external interfacing.

2) To have compatible software available that can handle

moderately sophisticated tasks.

3) To be S-100 bus compatible.

4) To be based on a widely utilized microprocessor.

5) To have a cathode ray tube (CRT) terminal to facilitate

communication and programming.

6) To have the minimum cost consistent with 1 through 5 above.

The Sol 20 microcomputer, manufactured by Processor Technology

Incorporated was selected. It offers an Intel 8080 microprocessor

based microcomputer, uses the S-100 bus and provides both parallel and

serial standardized port interfaces with connectors, and an audio

cassette interface capable of controlling two recorders. It includes

a monitor program capable of operating programs written directly in









machine code. The microcomputer includes a serially encoded ASCII

keyboard and CRT that are useful not only in programming the micro-

computer but also in utilizing the CRT terminal as a remote terminal

(via an acoustical coupler modem) to the local AMDAHL 470 system or

other compatible system. For more sophisticated programming require-

ments, data can be transferred from the microcomputer memory to the

AMDAHL 470 and vice versa. A BASIC (Beginners All Purpose Symbolic

Instruction Code) program language that can be loaded in random access

memory (requiring approximately 5000 bytes of memory) is included. The

BASIC language has sufficient computing capability to operate the camera

built in this research. It facilitates transfer of program control

between BASIC to machine code programming.

A BASIC program was developed that allowed data from the parallel

port to be entered into memory. Each of the detectors was associated

with a particular memory location. Each time that data were entered from

the parallel port, that particular memory location was incremented and

the resultant sum examined to see whether a predetermined maximum count had

been achieved in that channel. The program was designed such that when

any one of the channels reached some predetermined count, data collection

would terminate. The program also included the capability of collecting

data from a flood or plane source and storing that in a separate memory

area for use in correcting for the individual sensitivities of the

detectors. Under program control the image memory or calibration memory

may be initialized (set to zero). The calibration matrix can be inverted

and multiplied by the image matrix, to develop a corrected image matrix.

The numerical elements of the calibration matrix and the corrected image

matrix can be displayed on the CRT.







The program also divides the image array into sixteen levels of

gray (coded 0 through 15) and formats the data in random access memory

for eventual use by a videographics interface. The sixteen gray levels

are derived by defining sixteen levels between 0 counts and the maximum

number of corrected counts per channel and assigning each matrix element

a value of 0 through 15 based on its corrected value.


Videographics Interface


The "Dazzler" videographics interface, manufactured by Cromemo

Incorporated, was utilized since it is the only interface readily avail-

able that will read the random access memory of a S-100 compatible device

and display the data (properly formatted) in a 64 x 64 matrix with 16

shades of gray at each matrix location. The Dazzler does not require

servicing by the microprocessor during operation.

The Dazzler requires only 2048 bytes of computer memory. The

computer memory is scanned using direct memory access. The data in the

computer memory are formatted onto a TV screen to give a 64 by 64 matrix

with 16 shades of gray available at each matrix location. The Dazzler

output is a standard composite video signal that goes directly to a

monitor or television video amplifier. The microprocessor does encounter

some wait states caused by the Dazzler (while it reads the memory),

but this does not interfere with accurate program execution.

"In displaying the pictures, a compromise should be made between

contrast and gradation. Contrast should not be amplified far beyond

the point at which statistical variations become visible" (Anger, 1967,

p. 542). Gradation refers to the number of different gray tones used









and contrast refers to the ability to visibly distinguish between two

gray tones. Sixteen shades of gray are more than adequate in this

application, since brightness and contrast controls on the CRT may be

varied to derive the most meaningful image.


Summary

The camera described below was designed and constructed.

Twenty-five CdTe detectors were mounted on a printed circuit board

to form a 5 x 5 matrix. Each detector is located within an individual

hole of a collimator composed of individual square tubes of tantalum.

The collimator was designed for use with Tc-99m photons. The X and Y

outputs of the matrix are amplified in a one-stage amplifier and sub-

sequently examined in a pulse height threshold device in an analog to

digital converter. Events representative of one X and one Y event

occurring within some coincident time frame, as judged by the digital

circuitry, are available for input to a microcomputer. A microcomputer

controls the input data, provides for image data collection, correction

and subsequent display via a videographics interface. The resultant

display is a 5 x 5 image consisting of variations of 16 shades of gray.














CHAPTER 5

RESULTS


The performance of the camera is best illustrated by images of

several pattern sources. Appropriate pattern source images, as recorded

by polaroid photographs of the video monitor, are presented in subsequent

figures. For comparison, images obtained with a conventional Anger

camera (Ohio Nuclear Series 100 Anger Camera) are also illustrated on

some of the figures.

A calibration matrix (a matrix of calibration factors obtained by

imaging a uniform plane source) is as shown in figure 15. The wide

variation evident in the matrix is discussed in the next chapter.

The resolution of the camera is illustrated for a point source in

figure 16. Figure 17 also demonstrates the resolution of the camera.

The two parallel lines are oriented along two rows of detectors with the

centers of the lines roughly aligned with the centerlines of the

detector rows. The resolution is essentially perfect in this configura-

tion since adjacent rows represent the extremes of the gray scale.

Figure 18 is a representation of the same lines rotated through approximately

45. The lines are centered around the central axis of the detector

matrix.

Figures 19 and 20 represent the limit of resolution for the matrix

constructed in this research. Figures 19 represents two equal strength

line sources that are just slightly further apart (3.5 mm) than the

physical dimension of one collimating tube (2.9 mm). Note that the










SOURCE: UNIFORM PLANE (FLOOD)


CALIBRATION DATA MATRIX

XI X2 X) X X5
Y1 2.00 1.34 6.29 2.48 4.57
Y2 1.77 1.00 1.57 1.33 1.30
Y3 1.35 1.68 1.98 1.77 4096.00*
Y 1.27 1.16 1.23 1.48 1.36
Y5 1.29 2.11 2.71 1.53 2.65


*DATA NOT APPLICABLE:DETECTOR REMOVED DUE TO NOISE


FIGURE 15 CALIBRATION DATA MATRIX























EXPOSURE CONDITIONS:
SOURCE: TC-99M POINT SOURCE (1MM DIAMETER)
GEOMETRY: CLOSE TO COLLIMATOR
SCALES: ARBITRARY AND UNEQUAL


XY_ DTC AMERA
-POINT CENTERED
OVER ONE DETECTOR


ANGER CAMERA
-HIGH RESOLUTION
COLLIMATOR (140 KEV)


POINT SOURCE IMAGES


FIGURE 16


















EXPOSURE CONDITIONS
SOURCE: TC-99M PARALLEL LINE SOURCE (1MM DIAMETER,
6MM APART)
GEOMETRY: CLOSE TO COLLIMATOR
SCALES: ARBITRARY AND UNEQUAL
ACTIVITIES: LINES HAVE EQUAL SPECIFIC ACTIVITIES


XY CDTE CAMERA
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA


ANGER CAMERA
-HIGH RESOLUTION
COLLIMATOR (140 KEV)


PARALLEL LINE SOURCE IMAGES (6MM-ALIGNED)


FIGURE 17

















EXPOSURE CONDITIONS:
SOURCE: TC-99M PARALLEL LINE SOURCE (1MM DIAMETER,
6MM APART)
GEOMETRY: CLOSE TO COLLIMATOR
SCALE: ARBITRARY
ACTIVITIES: LINES HAVE EQUAL SPECIFIC ACTIVITIES






rj.




XY CDTE CAMERA
-LINES ROTATED 450
W.R.T, ROWS
-LINE LENGTH DOES NOT EXCEED
SENSITIVE AREA


PARALLEL LINE SOURCE IMAGE (6MM-ROTATED)


FIGURE 18

















EXPOSURE CONDITIONS:
SOURCE: Tc-99M PARALLEL LINE SOURCE (1MM DIAMETER,
3,5MM APART)
GEOMETRY: CLOSE TO COLLIMATOR
SCALES: ARBITRARY AND UNEQUAL
ACTIVITIES: LINES HAVE EQUAL SPECIFIC ACTIVITIES


I


XT CDTE CAMERA
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE AREA


ANGER CAMERA
-HIGH RESOLUTION
COLLIMATOR (140 KEV)


PARALLEL LINE SOURCE IMAGES (3.5MM-ALIGNED)


SI,iuPE 19


















EXPOSURE CONDITIONS:
SOURCE: TC-99M PARALLEL LINE SOURCE (1MM DIAMETER,
2,5MM APART)
GEOMETRY: CLOSE TO COLLIMATOR
SCALES: ARBITRARY AND UNEQUAL
ACTIVITIES: SPECIFIC ACTIVITY OF ONE LINE IS TWICE
THAT OF THE OTHER LINE





I f1


XY CDTE CAMERA
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE AREA


ANGER CAMERA
-HIGH RESOLUTION
COLLIMATOR (140 KEV)


PARALLEL LINE SOURCE IMAGES (2.5MM-ALIGNED)


FIGURE 20










center column does show some activity since the edges of both lines

overlap into the center column. Figure 20 illustrates that two line

sources of different activities can be imaged although the distance

between the lines (2.5 mm) is less than the physical dimension of one

collimating tube if the lines are oriented over two adjacent columns.

Degradation of image quality is observed by the introduction of 0.43

inch and 0.93 inch of Plexiglass scattering media between the lines

and the detector matrix is illustrated in figure 21 and figure 22. As

discussed subsequently, scattered radiation was not accurately discriminated

against. Geometry also contributes to the image degradation.

It was interesting to note during use of parallel line sources,

that air bubbles created within the device when the lines were loaded

could be diagnosed by the camera as shown in figure 23.

The sensitivity of the camera is qualitatively illustrated in

figure 24. The image was acquired in a counting time of approximately

3.8 seconds.

To illustrate the ability of the camera to image volume sources,

a thyroid phantom was imaged. The thyroid image represented in figure

25 is actually the composite of 36 images. Each image was normalized to

a common time frame by correcting for counting time and for decay of

the radioactivity. The 5 mm cold spot in the upper left hand corner is

somewhat spread out. This occurred because the cold spot was located at

the juncture of four composite images. Subsequent images of the cold

spot, with the cold spot centered over the detector matrix produced a

much sharper and better defined image.

The response of the camera was checked for paralysis by utilizing

a variety of uniform plane source activities (to a maximum of 0.146










mCi/cm2 of 99mTc). The counting rate was directly proportional to

source activity and no paralysis was evident.

The coincident time period was varied from near 0 to 5 seconds

with no apparent change in image quality.

The background count rate was so low as to be inconsequential.

The X Y CdTe camera had substantial imaging capabilities as il-

lustrated in the preceding figures. The results achieved as well as

difficulties encountered in achieving those results are discussed in

the next chapter.

















EXPOSURE CONDITIONS:
SOURCE: TC-99M PARALLEL LINE SOURCE (1MM DIAMETER,
6MM APART)
GEOMETRY: 0,43 INCHES OF PLEXIGLAS SCATTERING MEDIA
IMPOSED BETWEEN COLLIMATOR AND SOURCE
SCALE: ARBITRARY
ACTIVITIES: LINES HAVE EQUAL SPECIFIC ACTIVITIES











XY CDTE CAMERA
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA


FIGURE 21


PARALLEL LINE SOURCE IMAGE (0.43 INCHES
OF SCATTERING MEDIA)

















EXPOSURE CONDITIONS:
SOURCE: TC-99M PARALLEL LINE SOURCE (1MM DIAMETER,
6MM APART)
GEOMETRY: 0.93 INCHES OF PLEXIGLAS SCATTERING MEDIA
IMPOSED BETWEEN COLLIMATOR AND SOURCE
SCALE: ARBITRARY
ACTIVITIES: LINES HAVE EQUAL SPECIFIC ACTIVITIES





K



XY CDTE CAMERA
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA


FIGURE 22


PARALLEL LINE SOURCE IMAGE (0,93 INCHES
OF SCATTERING MEDIA)





























X J UR-E 011) IT AP I










L I J L ( L L L T




















-I t

1 1J2V
l,;ir cKT

':O'I IF A~.ilE r cO~;~u















EXPOSURE CONDITIONS:
SOURCE: TC-99M PARALLEL LINE SOURCE IMM DIAMETER,
5.5MM APART)
GEOMETRY: CLOSE TO COLLIMATOR
SCALE: ARBITRARY
ACTIVITIES: LEFT LINE-68$CI
RIGHT LINE-138CI
IMAGING TIME: 3.8 SECONDS





I


XY CDTE CAMERA
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE VOLUME


PARALLEL LINE SOURCE IMAGE (3.8 SECONDS)


FIGURE 24





















EXPOSURE CONDITIONS:
SOURCE: PICKER THYROID PHANTOM (Tc-99M)
GEOMETRY: CLOSE TO COLLIMATOR
SCALES: ARBITRARY AND UNEQUAL


XY CDTE CAMERA
-COMPOSITE OF 36
IMAGES


ANGER CAMERA
-HIGH RESOLUTION
COLLIMATOR (140 KEV)
-PINHOLE COLLIMATOR GAVE
EQUIVALENT RESULTS


THYROID PHANTOM IMAGES


FIGURE 25














CHAPTER 6

DISCUSSION OF RESULTS


The X Y CdTe camera concept offers the advantage of direct digital

localization of detected events, improvement of pulse height resolution

and attenuation of scatter within the detecting medium. As shown in the

previous chapter, the X Y CdTe camera produced images superior to other

existing devices. The images indicate that line source detection can

essentially be limited to the detectors directly beneath the line and

that imaging time is shorter than the other cameras tested. The per-

formance of the X Y CdTe camera, however, was not without limitations.

The performance of the camera was limited by two important problems:

1) The photopeak pulse heights of the individual CdTe detectors

varied by a factor of approximately two. The detector rows

were grouped according to pulse heights (i.e., each row had

detectors with similar pulse heights); however, as a result,

each detector column had a spectrum of photopeak pulse heights

ranging from very low to very high. Pulse height discrimina-

tion for all columns was established at a level low enough

to accommodate detection of events in the detectors with very

small pulse heights. As a result, the detectors having

larger pulse heights produced a lot more counts than detectors

having small pulse heights.

2) Although at one time there were 25 detectors that exhibited

photopeaks, after a relatively short usage time, several of










the detectors became noisy when operated at a field strength

of 325 V/cm. In order to utilize the detector matrix the

bias voltage had to be reduced to 150 V/cm which precluded

spectrometry. When operated below 325 V/cm, the signal to

noise ratio is greatly reduced (particularly for those

detectors exhibiting small photopeak pulse heights initially)

and photopeak resolution capabilities are lost. The reason

for the noisy performance is essentially undetermined.

Throughout the research, a number of CdTe detectors failed

(exhibited no photopeak or became excessively noisy). The

crystal surfaces exposed in cutting were not lapped or

polished after cutting but then the original crystals were

not lapped and polished on all sides either (particularly the

ends of the crystals). The surface preparation done on the

original crystals was done by hand and was rather crude.

The effect of the two problems discussed above was that there was

a very large variation in detector sensitivity that could not be

balanced out by adjusting discrimination levels and elimination of

scatter events by pulse height discrimination was not possible (which made

elimination of scatter by the placement of detectors within the col-

limating holes even more important).

The wide variation in detector sensitivities made conventional

analysis of plane sensitivity, modulation transfer function, line spread

function, etc., impractical due to the rather large statistical varia-

tion in the less sensitive detectors. The variation in detector

sensitivities is illustrated in figure 15. The ability to measure line

spread function wasn't crucial; however, since as shown previously the










full width at half maximum is essentially the width of one matrix

element, and the resolving capabilities of the camera are rather easily

illustrated.

The performance of the camera was compared to the rotational

parallel plate imaging device (PPRID) designed and constructed at the

University of Florida (Luthmann, 1978). Although the ultimate sensitivity

possible with either device has not been determined, it is possible to

establish that the PPRID has superior sensitivity to a conventional

Anger camera (see figure 26) since it can resolve two parallel lines of

75 pCi each in 9.4 sec. As shown in figure 24, the camera developed in

this research can resolve two lines of 138 pCi and 68 pCi in only 3.8

sec. Although the ability of both cameras to resolve two parallel lines

appears to be about the same, the resolution of the camera described

herein is somewhat superior in that the transition from the no-activity

regions to full-activity regions is absolute (i.e., goes immediately

and directly from the lightest gray scale to the darkest gray scale).

The PPRID camera produces a multitude of gray levels around and between

the two lines (not evident in figure 26 due to photographic technique),

such that the full width at half maximum encompasses more than the width

of one matrix element. Although both cameras seem to have about the

same ability to resolve parallel lines, thyroid phantom cold spots, etc.,

in all cases the image is somewhat superior for the camera described

herein since the contrast is superior. This would be expected since

the PPRID depends upon a computed reconstruction of the image from a

multitude of related images. The PPRID camera also has positioning

errors associated with it that do not have a direct analogy in this

camera. The radiologic image is directly measured by the X Y CdTe camera

described herein and hence should produce better images.




















Front Side Lucite

1.5mm 01mm
L- _

5mm 25mm

DUAL LINE SOURCE
75 Ci / Line


ROTATING LAMINAR CAMERA


.5 mm Imm Lucite
mmFront Side
Hole

U Li Lj
5 5 mm 12mm

TRIPLE LINE SOURCE
27uCi/Line


ANGER CAMERA


Condition: Radioactive Lines at Face of Collimator


Measuring time: 9.4 sec.
25 Profiles; 3/8 sec/Profile
Total counts: 9525







Measuring time: 9.4 sec.
25 Profiles; 3/8 sec./Profile
Total counts: 6401


Measuring time: 18.9 sec.
Total counts: 10,000








Measuring time: 15.6 sec.
Total count : 5000


PPRID AND ANGER CAMERA COMPARISONS


FIGURE 26






73



In summary, the X Y CdTe camera developed excellent resolution

and sensitivity. The direct digital localization was primarily

responsible for the success; however, in the absence of pulse height

resolution capabilities, the attenuation of scatter between elements in

the detector matrix became more important.














CHAPTER 7

CLINICAL FEASIBILITY


A clinically significant device should have a large enough sensitive

area to perform large organ studies or be large enough to be scanned or

composite to assimilate large organ images. Anger cameras have sensitive
2
areas as large as approximately 1000 cm Some of the considerations

for a clinical device are discussed below.

1) The matrix of a clinical device could be composed of sub-

groups. This is dependent upon the number (or volume) of

CdTe detectors that can be operated in a row without

serious degradation of signal to noise ratio or pulse

height resolution. During testing of the small crystals

after cutting operations, it was demonstrated that at least

five of the small crystals could be operated in a roworcolumn

while still maintaining a distinguishable photopeak. The

manufacturer of CdTe has demonstrated that four to eight

detectors for a total aggregate volume of approximately

2 x 2 x 50 mm can be operated and still maintain a dis-

tinguishable photopeak. The photopeak resolution of the

aggregate is substantially worse than any of the individual

detectors due to statistical signal to noise degradation

and due to the fact that no two detectors have essentially

the same photopeak pulse height which requires that the

detectors be roughly grouped according to pulse height










magnitude and according to noise. It should be noted

that the noise of the original crystals available varied

from 25 to 200 nA.

2) The size of a subgroup would essentially be determined by

the properties of the detectors available (i.e., limited

by noise and photopeak resolution considerations for

crystals in rows). A microcomputer system with two 8-bit

parallel input ports could easily handle a 256 x 256 matrix.

A 256 x 256 matrix would be adequate for clinical devices (even

at 1 mm x 1 mm element dimensions).

3) The size of the crystals and associated, insulation and

shielding between detectors could be reduced to produce

finer resolution. The ultimate size limitation depends upon

the ability of a manufacturer to mass produce small

crystals with consistent detection properties. A reduction

in the size of a matrix element requires an attendant increase

in associated electronics.

4) The crystals could ultimately be mounted directly on a

silicon substrate which could facilitate alignment of the

crystals, handling of large numbers of crystals and lead wire

attachment. Orientation could be established such that the

negative electrode is irradiated so as to take full advantage

of electron mobility (in comparison to hole mobility).

5) The concept of placing attenuating media between detectors

could be altered such that narrow strips of shielding could

be placed between the detectors rather than inserting the

detectors into the actual collimator tubes. This would enable










the ability to use several collimator designs (as is done

on the Anger camera) without sacrificing the absorption of

scatter radiation between detectors.

6) The analog portion of the circuitry could be redesigned to

make the voltage output independent of the position of charge

deposition (and hence increase the proportion of detections

indicative of full energy photon detection). By utilizing

a circuit time constant that is much much shorter than the

collection time for electrons, the slope of the output

pulse (dV/dt) would be proportional to the number of ion

pairs produced and the magnitude of the output becomes:

V = R(dq/dt).

The output voltage pulse is proportional to the current flow

and hence its height is a measure of the number of ion pairs

created (Price, 1964). The closed loop gain of an amplifier

with such a short circuit time constant would be very low and

hence a second amplifier stage might be required. Design

considerations for low noise second stage amplifiers are

available (Arbel, 1968). Alternately, a precision microvolt

comparator could be used. The use of a very short circuit

time constant would facilitate higher sensitivity (higher

count rate) since pulse duration would be shorter. The

performance of the detectors and the performance of the as-

sociated analog circuitry could be improved by cryogenic

operation (Elad, 1969).

7) Means for adjusting the timing of each analog to digital

electronics circuit would be useful in order to synchronize

the response to a given input pulse.










8) High speed TTL FIFOs are currently available that input

clock rate capabilities in excess of 10 MHZ. By ganging

large numbers of these relatively inexpensive memories, it

would be possible to store data as fast as the detector

matrix produces it and hence maximize camera sensitivity.

The data could later be transferred into computer memory at

the much slower rate required for that data transfer.

9) The image might be improved by inserting interpolated data

between actual detector measured data, based on the average

of measured data from adjacent elements. This could tend to

smooth out the transition of data between measured data

elements. An alternate gray level calculation scheme could

also be used. Rather than simply identifying sixteen levels

between zero and the maximum count obtained, it is possible

to define sixteen gray levels between the minimum count and

maximum count or base the image around an average count.

These procedures will not alter the basic data upon which the

image is based, but may produce an image that is aesthetically

more pleasing or easier to visualize. It should also be noted

that color images can also be produced.

The basic clinical feasibility ofa camera based on the

concept of constructing a large matrix of small detectors depends

upon the availability of large numbers of small detectors that

have essentially the same radiation detection properties. The

variability that exists, even in specially selected or grouped

CdTe detectors, is too large to fabricate a CdTe matrix camera










at this time. The recent availability of intrinsic germanium

detectors (1 mm x 1 mm x 7 mm) might make a large camera

feasible. However, the uniformity of radiation detection

properties and the pulse height resolution of a large series

of the detectors is unknown.

The direct digital localization of detected events and

the attenuation of scatter radiation in the detector plane

offer great promise for clinical devices. The availability

of large numbers of small detectors with uniform detection

characteristics would make the camera clinically feasible.














BIBLIOGRAPHY


Allemand R, Bouteiller P, Laval M: Present limitations of CdTe
detectors in nuclear medicine. Rev Phys Appl 12:365-367, 1977

Anger HO: Radioisotope cameras. Instrumentation in Nuclear Medicine.
New York, Academic Press, 1967, pp 485-552

Anger HO: Instruments: specific devices. Nuclear Medicine. New York,
HP Publishing Co, 1975, pp 29-40

Arbel FA: The second stage noise contribution of a nuclear pulse
amplifier. IEEE Trans Nucl Sci 15:2-5, 1968

Armantrout GA, Swierkowski SP,.Sherohman JW, Yee JH: What can be
expected from high z semiconductor detectors? IEEE Trans Nucl Sci
24:121-125, 1977

Aven M, Prener JS, Editors: Physics and Chemistry of II-VI Compounds.
Amsterdam, North Holland Publishing Co, 1967

Beck RN: Instruments: basic principles. Nuclear Medicine. New York,
HP Publishing Co, 1975, pp 17-28

Causer DA: The design of parallel hole gamma camera collimators. Int J
Appl Radiat Iso 26:355-362, 1975

Detko JF: A prototype ultra pure germanium orthogonal strip gamma
camera. Med Radioisotopes Scintigraphy 1:241-254, 1972

Elad E: FET preamplifiers for semiconductor radiation detectors.
ISPRA Nuclear Electronics Symposium Proceedings, 1969 pp 21-34

Entine G, Garcia DA, Tow DE: Review of CdTe medical applications. Rev
Phys Appl 12:355-359, 1977

ESPI Catalog 690. Los Angeles, Electronic Space Products Incorporated
1969, p 181

Hoffer PB, Beck RN, Gottschalk A, Editors: Semiconductor Detectors in the
Future of Nuclear Medicine. New York, The Society of Nuclear
Medicine, 1971, p 287

Hubbell JH: Photon Cross Sections, Attenuation Coefficients, and Energy
Absorption Coefficient From 10 keV to 100 GeV. Washington,
National Bureau of Standards, 1969, NSRDS-NBS29










Luthmann RW: Rotational parallel plate radionuclide imaging device.
Doctoral Dissertation, University of Florida, 1978

Malm HL, Raudorf TW, Martini M, Zanio KR: Gamma ray efficiency
comparisons for Si(Li), Ge, CdTe and Hgl2 detectors. IEEE Trans
Nucl Sci 20:500-509, 1973

Meyer E, Desaulniers G, Sternberg J: The CdTe subminiature semiconductor
detector: a versatile instrument for detection of gamma radiation.
Int J Appl Radiat Iso 26:697-702, 1975

Moretti JL, Mensch B, Guey A, Desgrey A: Comparative Assays of Scintil-
lation Cameras (Anger Type). Cleveland, Nuclear Medicine Institute,
1974

Myhill J, Hine GJ: Multihole collimators for scanning. Instrumentation
in Nuclear Medicine, New York, Academic Press, 1967, pp 429-460

Palms JM: Newer developments in detector design and materials.
Semiconductor Detectors in the Future of Nuclear Medicine.
New York, The Society of Nuclear Medicine Inc., 1971, pp 58-77

Price WJ: Nuclear Radiation Detection. New York, McGraw-Hill Book
Company, 1964

Serreze HB, Entine G, Bell RO, Wald FV: Advances in CdTe gamma ray
detectors. IEEE Trans Nucl Sci 21:404-407, 1974

Siffert P, Cornet A, Stuck R, Triboulet R, Marfaing Y: Cadmium telluride
nuclear radiation detectors. IEEE Trans Nucl Sci 22:211-225, 1975

Siffert P, Berger J, Scharager C, Cornet A, Stuck R, Bell RO,
Serreze HB, Wald FV: Polarization in cadmium telluride nuclear
radiation detectors. IEEE Trans Nucl Sci 23:159-170, 1976

Sveinsdottir E, Larsen B, Rommer P, Lassen NA: A multidetector scintil-
lation camera with 254 channels. J Nucl Med 18:168-174, 1977

Wald FV: Applications of CdTe, a review. Rev Phys Appl 12:277-290, 1977

Zanio K: Use of various device geometries to improve the performance
of CdTe detectors. Rev Phys Appl 12, 365-367, 1977














BIOGRAPHICAL SKETCH


Jerry David Allison was born on June 18, 1948 in Brevard, North

Carolina. He attended local primary and elementary schools and entered

North Carolina State University in 1966. He was graduated in 1970 with

a Bachelor of Science in nuclear engineering. Mr. Allison was employed

at Newport News Shipbuilding in Virginia as a Radiological Control Engineer

from 1970 until 1975. In 1974, he earned the Master of Engineering

degree in mechanical engineering from Old Dominion University in Norfolk.

In 1975, he entered the University of Florida for graduate work in

medical radiation physics.

Mr. Allison is a licensed Professional Engineer in the Commonwealth

of Virginia and is certified in health physics by the American Board

of Health Physics.

He is married to the former Mary Jacqueline Andrews, also of

Brevard. The Allisons have two sons.










I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.

I-


Walter Mauderli, Chairman
Professor of Nuclear Engineering
Sciences


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.




W. Emmett Bolch
Professor of Nuclear Engineering
Sciences


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.




Valerie A Brookeman
Associate Professor of Radiology


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.




Lawrence T. Fitzgerald
Assistant Professor of Nuclear
Engineering Sciences











I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.

/ 1 "

Genevieve S. Roessler
Assistant Professor of Nuclear
Engineering Sciences



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.


i 'X
Henri A. Van Rinsvelt
Professor of Physics



This dissertation was submitted to the Graduate Faculty of the College
of Engineering and to the Graduate Council, and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.

August 1978



Dean, College of Engineering


Dean, Graduate School






UF Libraries: Digital Dissertation Project


Dear Dr. Jerry D. Allison,

The George A. Smathers Libraries at the University of Florida has initiated a project to retrospectively digitize and
make available on the Internet any dissertation written by a University of Florida doctoral candidate and accepted by
the University of Florida. It is our goal to make the documents fully text searchable and easily harvested by Internet
search engines, allowing the full breadth and scope of scholarship produced at the University of Florida to be made
available across the world quickly and easily. The Library is bearing the full cost of the project.
We would like to add your dissertation, The application of cadmium telluride as in hole semiconductor radiation
detectors in an X Y matrix radioisotope camera /, published in 1978, to the project. In order to do so we need a
signed, legal original Internet Distribution Consent Agreement for our files. If you want your dissertation included in
the project, please print the Consent Agreement on the page below, sign it and mail it back to the Libraries at the
address listed.

We will keep you informed of the progress of your dissertation as it works its way through the project.

If you have any questions, please reply to UFdissertations@.uflib.ufl.edu.
Thank you,

Cathy Martyniak, Project Coordinator
Christy Shorey, Project Technician


FW UNIVERSITY of
UFLORIDA
The Foundation for The Gator Nation



Internet Distribution Consent Agreement

In reference to the following dissertation:

AUTHOR: Allison, Jerry
TITLE: The application of cadmium telluride as in hole semiconductor radiation
detectors in an X Y matrix radioisotope camera / (record number: 74791)
PUBLICATION DATE: 1978



I, 1 L 2 ? Il as copyright holder for the aforementioned dissertation, hereby grant
specifiea d limited ar e and distribution rights to the Board of Trustees of the University of Florida and its agents. I authorize
the li'ersity of Flori to digitize and distribute the dissertation described above for nonprofit, educational purposes via the
Internet or successive technologies.

This is a non-exclusive grant of permissions for specific off-line and on-line uses for an indefinite term. Off-line uses shall be
limited to those specifically allowed by "Fair Use" as prescribed by the terms of United States copyright legislation (cf, Title 17,
U.S. Code) as well as to the maintenance and preservation of a digital archive copy. Digitization allows the University of Florida
to generate image- and text-based versions as appropriate and to provide and enhance access using search software.
This gr a f permissions prohibb e h igitized versions for commercial use or profit.

e of Copyright Id

n Imain: Page 1 of 2


6/17/08 9:21. AM







UF Library: Digital Dissertation Project


Printed nr Tvned Na4ne nf to'nQiht Holder/Licensee A


Personal information blurred




Datef Siature

Please print, sign and return to:
Cathleen Martyniak
UF Dissertation Project
Preservation Department
University of Florida Libraries
P.O. Box 117007
Gainesville, FL 32611-7007


mhtmlmain:


6/17/08 9:21. AM


F / A


Page 2 of 2




Full Text
4
camera and is not really applicable to this research.
The application of CdTe to radioisotope cameras is
sufficiently diversified to assure that each study is
unique and significant.
2) Semiconductor detectors have not been employed in an inhole
matrix geometry. Cameras have been made that utilize
orthogonal strip germanium detectors (Detko, 1972). The
cameras have demonstrated significant resolution but are
somewhat limited in size. Fabrication of larger cameras from
several orthogonal strip detectors is underway. The
germanium requires cooling during operation and since each
orthogonal strip detector is actually one detector with many
leads, some cross talk between elements exists (particularly
when adjacent conducting strips are in close proximity for good
resolution). The orthogonal strip arrangement does not lend
itself to shielding between elements so as to reduce scattered
radiation contributions. The extension of the orthogonal
conducting strips beyond the ideal sensitive volume of the
detectors creates substantial nonuniform electric fields near
the edges of each detector. Within the regions of nonuniform
electric fields, the field strengths are less than in the
more sensitive regions. As a result, incomplete charge
collection occurs and substantial numbers of pulses with
reduced pulse height are produced that must be discriminated
against. The limitation on spacing between adjacent con
ducting strips creates substantial gaps between detectors


75
magnitude and according to noise. It should be noted
that the noise of the original crystals available varied
from 25 to 200 nA.
2) The size of a subgroup would essentially be determined by
the properties of the detectors available (i.e., limited
by noise and photopeak resolution considerations for
crystals in rows). A microcomputer system with two 8-bit
parallel input ports could easily handle a 256 x 256 matrix.
A 256 x 256 matrix would be adequate for clinical devices (even
at 1 mm x 1 mm element dimensions).
3) The size of the crystals and associated, insulation and
shielding between detectors could be reduced to produce
finer resolution. The ultimate size limitation depends upon
the ability of a manufacturer to mass produce small
crystals with consistent detection properties. A reduction
in the size of a matrix element requires an attendant increase
in associated electronics.
4) The crystals could ultimately be mounted directly on a
silicon substrate which could facilitate alignment of the
crystals, handling of large numbers of crystals and lead wire
attachment. Orientation could be established such that the
negative electrode is irradiated so as to take full advantage
of electron mobility (in comparison to hole mobility).
5) The concept of placing attenuating media between detectors
could be altered such that narrow strips of shielding could
be placed between the detectors rather than inserting the
detectors into the actual collimator tubes. This would enable


9
energy resolution. The energy resolution of a solid state detector
is about twenty times higher than that of a conventional Nal(Tl)
scintillation detector. In addition, the elimination of scattered
radiation contributes to the improvement of image quality. The semi
conductors are considerably more efficient than Geiger probes and their
bias voltage is lower than that required by any other probe.
At present, solid state detectors are being used in the imaging
techniques; however, thus far attempts to make a high resolution semi
conductor gamma camera met with considerable difficulties" (Meyer, 1975,
p. 697).
Gas ionization chambers can also be fabricated into relatively
small dimensions but are limited in usefulness by the relatively inef
ficient deposition of energy to the gas contained therein.
Attempts at utilizing semiconductor detectors in high resolution
gamma cameras have consisted thus far of placing the detector close to
the back surface of the collimator. The research described herein,
utilizes the semiconductor detectors within the holes of the collimator
device rather than in back of the collimator device.
It is important that the collimator be located as close as possible
in order to prevent photons scattered on septa surfaces from being
detected in adjacent detectors and producing counts that do not give
an accurate representation of the origin of the photon.
The in hole configuration utilizes the small size into which semi
conductor detectors can be fabricated. Nal(Tl) detectors can be
fabricated into relatively small detectors but their use is somewhat
limited by the requirement that they be hermetically sealed and the


78
at this time. The recent availability of intrinsic germanium
detectors (1 mm x 1 mm x 7 mm) might make a large camera
feasible. However, the uniformity of radiation detection
properties and the pulse height resolution of a large series
of the detectors is unknown.
The direct digital localization of detected events and
the attenuation of scatter radiation in the detector plane
offer great promise for clinical devices. The availability
of large numbers of small detectors with uniform detection
characteristics would make the camera clinically feasible.


30
Figure 7 Printed Circuit Board


CHAPTER 3
CAMERA PROPERTIES
Direct digital localization of detected events in the X Y CdTe
camera developed in this research is an advantage in comparison to the
Anger camera. Also, the utilization of semiconductor radiation detectors
(particularly CdTe) and the placement of the detectors within collimating
holes offer some advantages over conventional radioisotope cameras.
To better appreciate the advantages, some of the properties of the
Anger camera and the X Y CdTe camera will be examined.
The deposition of energy, by radiation, within the Nal(Tl) crystal
of an Anger camera results in scintillations proportional to the ioniza
tion and excitation produced. The scintillations are detected by an
array of PMTs which convert the light into electrical signals proportional
in amplitude to the quantity of light incident upon the PMTs. The
location of the energy deposition is determined via a capacitive net
work that divides the pulse amplitudes according to the distribution of
luminescence with respect to the positions of the PMTs. The resultant
pulses are then converted to visible light flashes at appropriate sites
on a cathode ray tube (CRT). The position of the flashes on the CRT
are determined by difference circuits or ratio circuits.
For purposes of forming an image of the distribution of radio
activity within an object, only photons created within the object that
reach the sensitive volume of the detector without scattering (coherent
7


PACKAGE
DESTINATION
DESOtlPTIOH
A.B.C.D.E.F.Y.HH
CD 4098
Dual one shot
C.l.D.V
CD 4072
Dual 4 Input OK gate
I.J.K.FF.GC
CD 4042
Quad clocked D latch
L.M.AA.BB
CD 4050
Hex buffer
N.T
CD 4078
8 input NOR gate
O.P.Q.R
CD 4081
Quad 2 input AND gate
S
CD 4011
Quad 2 input NAND gate
w.x
CD 4082
Dual 4 input AND gate
CC.DD
3341
64 word-4 bit first in first
out serial wewory(FIFO)
EE
SN 7401
Quad 2 input NAND gate
-R-
IN3064
Diode
: 14
Digital
Circuitry (Page 2
OF 2)


ACKNOWLEDGMENTS
The members of my advisory committee (Drs. W. Emmett Bolch, Valerie
A. Brookeman, Lawrence T. Fitzgerald, Walter Mauderli, Genevieve S.
Roessler and Henri A. Van Rinsvelt) were instrumental in the formula
tion, accomplishment and documentation of this research. Dr. Mauderli
served as chairman of the committee and provided special assistance
with analog electronics. Dr. Fitzgerald provided special assistance
with digital electronics and microcomputer applications. Dr. Bolch,
through the Department of Environmental Engineering Sciences, provided
a graduate assistantship during my first year at the University.
Dr. Genevieve Roessler counseled me with regard to course work and
helped me find appropriate assistantships throughout my graduate study
at the University.
Dr. Clyde M. Williams, as Chairman of the Department of Radiology,
provided a graduate assistantship for myself and financial support for
my research project.
The person most responsible for making this dissertation possible
is my wife, Jacquie. Jacquie and I left rather secure professional
careers and a very pleasant social environment in Newport News,
Virginia, so that I could pursue a doctorate at the University of Florida.
Not only did Jacquie provide moral support but also supported our family
financially by teaching in the secondary schools of Alachua County.
She bore our second child and completed her master's degree during our
three year stay in Gainesville. Jacquie made this dissertation possible.
iii


46
Digital Circuitry
The digital circuitry was developed to meet the following criteria:
1) To identify, based on the digital outputs of the row
and column circuitry the individual detector in which
a detection event occurs and to make that data
available for input to a microcomputer.
2) To eliminate from consideration, those events in which
more than one row signal or more than one column signal,
or both, occur within some coincident time frame.
The digital circuitry was designed such that the only events that
are collected as data are those in which one row signal and one column
signal occurred simultaneously (within some coincident time period) and
hence identify that a detection event occurred in a particular detector
(see figure 14). When a data input line goes high (after some quiescent
period when all inputs are low), the circuitry (after a brief wait for
coincidence), latches the data present. If the data consist of one X
signal and one Y signal, then that data are entered into a storage
device for subsequent transfer to a microcomputer port. If the data
are not acceptable or after completion of the data transfer, the process
starts again. The time frame in this particular device is not of
great importance since there are only 25 detectors supplying signals
and since the microcomputer requires a relatively long period of time
to input data.
The dual monostable multivibrators (one shots) A, B, C, D and E
serve to produce one 9.1 psec pulse each time that any of the XI X5
comparators or Y1 Y5comparators produce a transition from logic 0 to
logic 1 at the one shot input while the one shot output is low. The


17
amplitude of the signals produced can be analyzed by
pulse height discrimination to eliminate some of the events
caused by scattered photons.
2) The X Y CdTe camera will not produce distortion near the
edges of the matrix. The generation of X Y positional informa
tion by the capacitive network pulse divider of the Anger
camera is based on the center of luminous intensity incident
on the PMT array. For detections that result in sufficient
luminescence production to be indicative of full energy
photons, events are displayed on the CRT in positional rela
tion to the effective center of luminescence incident on the
PMTs. Due to inhomogeneities and defects in the Nal(Tl)
crystal and due to reflection of light at the crystal
boundaries, the center of luminescence incident on the PMTs
is not in the exact position of energy deposition by the
radiation. This distortion is relatively small in the
center of the crystal but increases near crystal edges. The
distortion is as high as 30% near the edges of some Anger
cameras (Moretti, 1974). Photons that undergo multiple
incoherent scatters within the Nal(Tl) crystal can cause
significant differences in the location of the center of
luminous intensity and the point of the photon's initial
interaction (which contains the positional information).
For a matrix of CdTe detectors, the location of a
photon detection is determined to be within a particular
detector via the X and Y address produced. The positional
information will not be distorted near the edge of the matrix


37
In keeping with these ideals, a one stage amplifier that has
sufficient amplification to convert a relatively small charge pulse
collected from the crystals (on the order of less than 0.005 picocoulombs
for 100 percent charge collection from a 140 keV photon) to a voltage
pulse that has sufficient magnitude to be used directly in an analog to
digital conversion device was developed (see figure 10). It has been
reported that the actual charge collection efficiency in CdTe is sub
stantially less than 100% and is dependent upon electric field strength
and energy deposition location (Malm, 1973). The amplifier is essentially
a charge sensitive amplifier or charge to voltage converter. A primary
advantage of the charge sensitive design is that the output is independent
of detector capacity (which may vary with time and number of detectors
per row). The detector is alternating current (a.c.) coupled to the
preamplifier in order to remove the direct current (d.c.) component prior
to amplification. Direct current coupling would be preferred since it
is somewhat less noisy, but in this application, signals are acquired
from both of the detector electrodes which requires that at least one
side be a.c. coupled in order to remove the bias voltage component
applied to the detector. The amplifier consists of an operational
amplifier with a junction field effect transistor (JFET) at the input.
The resultant voltage output (V) is related to the charge input (Q) by
the expression:
V = Cf
where represents the capacity between the input and output of the
amplifier. In order to maximize the voltage output for a given input,
the capacity should be minimized, thus no added capacity was utilized


68
Exposure Conditions:
Source: Picker thyroid phantom (Tc-99m)
Geometry: close to collimator
Scales: arbitrary and unequal
XY CdTe Camera
Anger Camera
-composite of 36 -high resolution
images collimator (190 keV)
-pinhole collimator gave
equivalent results
Figure 25 Thyroid Phantom Images


CHAPTER 1
INTRODUCTION
Radioactive tracer studies are an integral part of modern medical
diagnostic procedures. By administering radioactive tracers (radio
pharmaceuticals) that are specific for some physiological function,
physicians are able to determine the level of that function within a
patient in terms of physical and chemical phenomena and hence determine
the site and rate of important biological processes. Radioisotope
cameras are the instruments that enable many of the relatively nonin-
vasive radioactive tracer studies that are currently prevalent in nuclear
medicine. It is because of the fact that improvements in spatial or
temporal resolution of radioisotope cameras have the potential of making
better diagnostic information available to the physician as well as
reducing radiation dose to the patient, that the research described
herein was undertaken and is important.
Radioisotope cameras image the distribution of radioisotopes in
vivo. A camera creates a two dimensional image of the three dimensional
radioisotope distribution within an object. Radioisotope cameras are
mainly of the gamma ray scintillation type (Anger camera). Anger
cameras employ a sodium iodide [Nal(Tl)] scintillation crystal of 1/8
to 1 inch thickness as the detecting medium. The Nal(Tl) crystal is
located in back of a collimator which creates a spatial relationship
between the image and the object.
1


CdTe
Dtector
Matrix
Figure 3 Block Diagram Of X Y CdTe Camera


150k
45
For Positive Analog Input Signals
(Voltage supplies are filtered as shown on figure 10.)
Analog
Input
For Negative Analog Input Signals
(Voltage supplies are filtered as shown on figure 10.)
Figure 13
Analog To Digital Converters


70
the detectors became noisy when operated at a field strength
of 325 V/cm. In order to utilize the detector matrix the
bias voltage had to be reduced to 150 V/cm which precluded
spectrometry. When operated below 325 V/cm, the signal to
noise ratio is greatly reduced (particularly for those
detectors exhibiting small photopeak pulse heights initially)
and photopeak resolution capabilities are lost. The reason
for the noisy performance is essentially undetermined.
Throughout the research, a number of CdTe detectors failed
(exhibited no photopeak or became excessively noisy). The
crystal surfaces exposed in cutting were not lapped or
polished after cutting but then the original crystals were
not lapped and polished on all sides either (particularly the
ends of the crystals). The surface preparation done on the
original crystals was done by hand and was rather crude.
The effect of the two problems discussed above was that there was
a very large variation in detector sensitivity that could not be
balanced out by adjusting discrimination levels and elimination of
scatter events by pulse height discrimination was not possible (which made
elimination of scatter by the placement of detectors within the col
limating holes even more important).
The wide variation in detector sensitivities made conventional
analysis of plane sensitivity, modulation transfer function, line spread
function, etc., impractical due to the rather large statistical varia
tion in the less sensitive detectors. The variation in detector
sensitivities is illustrated in figure 15. The ability to measure line
spread function wasn't crucial; however, since as shown previously the


Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: close to collimator
Scale: arbitrary
Activities: lines have equal specific activities
XY CdTe Camera
-lines rotated A5
W.R.T. ROWS
-LINE LENGTH DOES NOT EXCEED
SENSITIVE AREA
Figure 18 Parallel Line Source Image (6mm-Rotated)


52
machine code. The microcomputer includes a serially encoded ASCII
keyboard and CRT that are useful not only in programming the micro
computer but also in utilizing the CRT terminal as a remote terminal
(via an acoustical coupler modem) to the local AMDAHL 470 system or
other compatible system. For more sophisticated programming require
ments, data can be transferred from the microcomputer memory to the
AMDAHL 470 and vice versa. A BASIC (Beginners All Purpose Symbolic
Instruction Code) program language that can be loaded in random access
memory (requiring approximately 5000 bytes of memory) is included. The
BASIC language has sufficient computing capability to operate the camera
built in this research. It facilitates transfer of program control
between BASIC to machine code programming.
A BASIC program was developed that allowed data from the parallel
port to be entered into memory. Each of the detectors was associated
with a particular memory location. Each time that data were entered from
the parallel port, that particular memory location was incremented and
the resultant sum examined to see whether a predetermined maximum count had
been achieved in that channel. The program was designed such that when
any one of the channels reached some predetermined count, data collection
would terminate. The program also included the capability of collecting
data from a flood or plane source and storing that in a separate memory
area for use in correcting for the individual sensitivities of the
detectors. Under program control the image memory or calibration memory
may be initialized (set to zero). The calibration matrix can be inverted
and multiplied by the image matrix, to develop a corrected image matrix.
The numerical elements of the calibration matrix and the corrected image
matrix can be displayed on the CRT.


5
that reduce the sensitive area of the matrix (in comparison
to total crystal area). Orthogonal strip cameras constructed
to date utilize large resistor divider networks to produce
one analog signal representing X position and one analog
signal representing Y position, rather than direct digital
localization of the detector in which a detection event
occurred.
3) Gamma cameras employing multiple Nal(Tl) crystals have not
been substantially successful in comparison to the Anger
camera. One such device, the autofluoroscope, was composed
of 294 Nal(Tl) crystals, each of which was coupled to an X
position and Y position photomultiplier tube (PMT) by long
light pipes. The autofluoroscope could operate at high count
rates but had poor photopeak energy resolution caused by
inefficient transfer of light through the light pipes (Beck,
1975). The poor energy resolution reduced the ability to
discriminate against scattered photons.
A camera, emoloying 254 Nal(Tl) crystals, was developed
recently (Sveinsdottir, 1977). The spatial resolution was
essentially the same as previous multidetector scintillation
cameras but the temporal resolution was substantially improved
by utilizing parallel event processing in lieu of sequential
event processing. The number of crystals used remains some
what limited by the use of one light pipe, one PMT, amplifier,
discriminator, buffer, etc. per crystal.
Another such device was the Quantascope camera which
employed 2515 cesium iodide CsI(Tl) crystals that were


Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
2.5mm apart)
Geometry: close to collimator
Scales: arbitrary and unequal
Activities: specific activity of one line is twice
THAT OF THE OTHER LINE
I
XY CdTe Camera
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE AREA

-HIGH RESOLUTION
COLLIMATOR (1A0 KEV)
Figure 20 Parallel Line Source Images (2.5mm-Aligned)


THE APPLICATION OF CADMIUM TELLURIDE AS IN HOLE SEMICONDUCTOR
RADIATION DETECTORS IN AN X Y MATRIX RADIOISOTOPE CAMERA
BY -
JERRY DAVID ALLISON
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE
UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1978


Exposure Conditions
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: close to collimator
Scales: arbitrary and unequal
Activities: lines have equal specific activities
XY CdTe Camera
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Anger Camera
-HIGH RESOLUT
COLLIMATOR
KEV)
Figure 17 Parallel Line Source Images (6mm-Aligned)


66
Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
5mm apart)
Geometry: close to collimatot
Scale: arbitrary
Activities: lines have equal specific activities (with
ONE AIR BUBBLE EVIDENT)
|
I
I
XY CdTe Camera
-LINES ALIGNED
WITH COLUMNS *BUBBLE (Air)
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 23 Parallel Line Source Image (Bubble Evident)


Figure 1 CdTe Detector Matrix
Bbiththtii
KFcPcFa-
tHtHta-ob
0500a
SffiB
O - I
cn
Vbias (-) 1.5N


51
The time frame for selection of a valid event is not very critical
in this application since the microcomputer uses in excess of 33 ysec
per data transfer. Complimentary metal oxide semiconductor (CMOS) devices
were utilized since they have low power consumption and good noise im
munity and since the data transfer time is not critical. Processing
time in the digital circuitry could be greatly reduced by employing
transistor-transistor logic (TTL) devices throughout and by incorporating
faster FIFOs.
Microcomputer
The microcomputer was selected to meet the following criteria:
1) To have sufficient hardware capabilities to readily input
data from a parallel port and store the data in memory with
minimal external interfacing.
2) To have compatible software available that can handle
moderately sophisticated tasks.
3) To be S-100 bus compatible.
4) To be based on a widely utilized microprocessor.
5) To have a cathode ray tube (CRT) terminal to facilitate
communication and programming.
6) To have the minimum cost consistent with 1 through 5 above.
The Sol 20 microcomputer, manufactured by Processor Technology
Incorporated was selected. It offers an Intel 8080 microprocessor
based microcomputer, uses the S-100 bus and provides both parallel and
serial standardized port interfaces with connectors, and an audio
cassette interface capable of controlling two recorders. It includes
a monitor program capable of operating programs written directly in


44
logic circuitry. The converter was also designed to facilitate hybridiza
tion. Since the output of amplifiers servicing rows is of opposite
polarity to the output of amplifiers servicing .columns, it was necessary
to develop converters for negative input signals and positive input
signals, that would produce a positive logic pulse when the input signal
exceeded some reference level.
The converters consist of voltage comparators as shown in figure 13.
The output of the comparators is fedback to the input in order to create
hysteresis and hence prevent input signal noise from causing the output
to switch states several times as the input signal passes through the
reference voltage level. Hysteresis serves to speed up the circuit
switching time and cause the comparator to switch only once as the input
signal passes through the reference voltage level and hence prevent
chatter. Chatter is deleterious to this application since it could
ultimately be evaluated as multiple detection events. The voltage
comparator was selected because it has a fast response time and reasonable
input offset and input bias currents. Capacitors were added from the
outputs of the comparators to ground in order to alleviate oscillations
caused by the output signal from one comparator being fedback by stray
coupling to the input of an adjacent comparator. The comparators were
assembled onto a printed circuit board but the inputs were inadequately
guarded to prevent the oscillations. The comparator printed circuit
board was shielded from the amplifier printed circuit board (and other
boards) in order to alleviate oscillation problems encountered due to
feedback caused by stray coupling.


23
photons is essentially restricted to within the physical
dimensions of the detector located in the collimator hole.
For a Nal(Tl) crystal located behind the collimator, a detec
tion could occur beneath an adjacent collimator hole.
Scattered photons (scattered on collimator surfaces or within
the detector) are significantly attenuated by the collimator
septa that separate individual detectors.
Pisadvantages
The application is not without some limitations. The disadvantages
are as follows:
1) A substantial gap will exist between adjacent detectors
resulting in a reduced detector area as compared to a solid
crystal such as an Anger camera employs. The gap produces a
sensitive detector area of approximately 50% of the overall
camera dimensions as compared to 65 to 80% for the Anger camera.
The sensitivity is therefore reduced somewhat (for equivalent
detector thicknesses). The ratio of the sensitive detector
area to the camera face area could be increased with more
sophisticated fabrication technology.
2) The theoretical limit of resolution is limited by centerline
to centerline spacing of the detectors. In this application,
the centerline to centerline distance may be in excess of 3 mm.
Should the application prove successful, the distance could be
reduced by smaller detectors and gaps.


Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: 0.93 inches of Plexiglas scattering media
IMPOSED BETWEEN COLLIMATOR AND SOURCE
Scale: arbitrary
Activities: lines have equal specific activities
XY CdTe Camera
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 22
Parallel Line Source Image (0.93 Inches
of Scattering Media)


80
Luthmann RW: Rotational parallel plate radionuclide imaging device.
Doctoral Dissertation, University of Florida, 1978
Malm HL, Raudorf TW, Martini M, Zanio KR: Gamma ray efficiency
comparisons for Si(Li), Ge, CdTe and Hgl0 detectors. IEEE Trans
Nucl Sci 20:500-509, 1973
Meyer E, Desaulniers G, Sternberg J: The CdTe subminiature semiconductor
detector: a versatile instrument for detection of gamma radiation.
Int J Appl Radiat Iso 26:697-702, 1975
Moretti JL, Mensch B, Guey A, Desgrey A: Comparative Assays of Scintil-
lation Cameras (Anger Type). Cleveland, Nuclear Medicine Institute,
1974
My hill J, Hie GJ: Multi hole collimators for scanning. Instrumentation
in Nuclear Medicine, New York, Academic Press, 1967, pp 429-460
Palms JM: Newer developments in detector design and materials.
Semiconductor Detectors in the Future of Nuclear Medicine.
New York, The Society of Nuclear Medicine Inc., 1971, pp 58-77
Price WJ: Nuclear Radiation Detection. New York, McGraw-Hill Book
Company, 1964
Serreze HB, Entine G, Bell R0, Wald FV: Advances in CdTe gamma ray
detectors. IEEE Trans Nucl Sci 21:404-407, 1974
Siffert P, Cornet A, Stuck R, Triboulet R, Marfaing Y: Cadmium telluride
nuclear radiation detectors. IEEE Trans Nucl Sci 22:211-225, 1975
Siffert P, Berger J, Scharager C, Cornet A, Stuck R, Bell R0,
Serreze HB, Wald FV: Polarization in cadmium telluride nuclear
radiation detectors. IEEE Trans Nucl Sci 23:159-170, 1976
Sveinsdottir E, Larsen B, Rommer P, Lassen NA: A multidetector scintil
lation camera with 254 channels. J Nucl Med 18:168-174, 1977
Wald FV: Applications of CdTe, a review. Rev Phys Appl 12:277-290, 1977
Zanio K: Use of various device geometries to improve the performance
of CdTe detectors. Rev Phys Appl 12, 365-367, 1977


39
between the input and output and the feedback capacity is the stray
capacitance between the input and output. A feedback resistor is
required to discharge the feedback capacity; otherwise, the feedback
capacity would build up a charge when input signals are present that
could eventually saturate the amplifier. A high value feedback resistor
is utilized in order to maintain a large time constant to maximize
the collection of charge deposited within the crystal. For the CdTe
detectors utilized in this research, an electric field of about 325 V/cm
gave optimum spectrometer performance. Since the electron and hole
2 2
mobilities of CdTe are on the order of 1000 cm /Vsec and 80 cm /Vsec
respectively, the charge collection time for a 325 V/cm field strength
is approximately 0.6 ysec for electrons and 8.0 ysec for holes. In
order to assure that the collection of positive charges (holes) and
negative charges (electrons) is maximized (in order to maximize the
amplifier output voltage per detection event), a 500 resistor was
utilized in the feedback network. The large feedback resistor thus assures
that the amplifier time constant is much much larger than the transit
time for hole collection and hence maximizes the charge collected per
detection event. The operational amplifier was selected for its high
gain and relatively low noise. The signals to be amplified occur at a
relatively low frequency (due to relatively long charge collection times
and due to a relatively low contemplated patient flux incident upon the
matrix), and hence the frequency-bandwidth characteristics of the
operational amplifier are not of primary importance. The JFET employed
at the input of the amplifier, determines the amplifier input impedance
and develops the signal/noise ratio that will be evident at the output
of the amplifier. The operational amplifier primarily provides the gain.


33
effective centers of the in hole detectors are located within the
collimator and hence existing design criteria are not exactly analagous.
Theoretical parallel multihole collimator designs attempt to optimize
the number of holes, their size, shape, septa thickness, and hole length.
Practical collimator design parameters have largely been assigned
empirically (My hill, 1967). Since the scope of this research was
primarily involved with the validity of the application rather than the
optimization of the collimator dimensions, selection of collimator dimen
sions was based on existing criteria for Anger type parallel hole col
limator design. Causer (1975) indicates that optimum collimator length
99m
is essentially independent of resolution and that for Tc photons,
optimum collimator length is approximately 1.8 2.5 cm. Collimator
length was selected as one inch since the effective collimator length
is reduced somewhat when the detector is located in hole. For a square
hole side dimension of about 0.115 inches, according to Anger (1967),
the septum thickness should be approximately 0.0081 inches. The actual
collimator dimensions are: length--1.0 inches (.01 inches); septa
thickness--0.008 inches ( 10%); square hole side dimension--0.115
inches ( 0.001 inches).
The shortest distance a photon can travel through a septum when
taking the unwanted path of minimum attenuation through the collimator
Q Q pi
is approximately 0.0355 inches. For Tc photons in tantalum this is
equivalent to an attenuation of 97%. The attenuation of scatter between
detectors for scattered photon energy of 140 keV is approximately 50%
and for scattered photon energy of 100 keV is approximately 80%.
The square tantalum tubes are assembled inside of an aluminum block
(see figure 8). Most of the tantalum tubes were larger than the square


11
its relatively high specific gravity (CdTe:5.20, Si:2.4, Ge:5.35). The
relatively high Z and specific gravity enhance the energy absorption
efficiency of CdTe.
The CdTe detectors used in this research are p type, chlorine doped
for compensation and have chemically deposited platinum electrodes.
Platinum is frequently used as the electrode material since its work
function is relatively high which helps to prevent polarization phenomenon
(Siffert, 1976).
The electron and hole mobilities are on the order of 1000 and
2
80 cm /V-sec, respectively. The electron and hole mobi1ity-trapping
-3 -4 2
time products are on the order of 10 and 10 cm /V respectively. The
detectors are characterized as homogeneous semiconductor radiation
detectors. The crystallographic orientation of the crystals to be used
is not precisely known. The crystals grow preferentially in the [111]
direction. Detector slices are cut perpendicularly to the direction of
growth and electrodes are applied across what is probably the [111]
principal axis. "The effect of crystallographic orientation of the
detector was examined for the applied electric bias along three
principal directions--[111], [110] and [100]--but no significant dif
ferences were observed" (Siffert, 1976, p. 159).
The pulse shape produced by the detectors is a function of the
position and orientation of charges liberated. Ideally the charge
liberation would be parallel to the electrodes and very near the negative
electrode, resulting in rapid, simultaneous collection of all charges.
If the charge liberation is perpendicular to the electrodes, the charges
nearest the electrodes are collected first and so forth, such that


26
Substantial work was done in developing cutting techniques to
minimize degradation of spectrometer performance during cutting. Noise,
photon energy spectra and stability with time were examined for each
large crystal prior to cutting and for the small cubes after cutting.
Figure 4 represents a typical energy spectrum for a large crystal prior
to cutting and one of the small cubes after cutting (the noise voltage
and leakage current are also represented).
Wire leads were attached to the platinum electrodes with conducting
silver epoxy. Attempts were first made at mechanically coupling a wire
spring to the electrodes, but the hardware required could not be easily
manufactured. Attempts to use a silicon rubber based conducting
compound were also futile since the resultant bond between the wire and
the platinum was not adequate to allow handling of the crystal by its
wire leads. Ultrasonic bonding was also investigated but disregarded
since it requires high local temperatures and does not work well on
platinum surfaces. The wire used in ultrasonic bonding would not be
strong enough to allow for handling the crystal by its wire leads.
The epoxy utilized is a two component silver filled epoxy designed
specifically for chip bonding in microelectronic and optoelectronic
applications. It cures at temperatures as low as 50C (for 12 hours).
A detector, prior to wire lead bonding, is shown in figure 5 and after
wire lead bonding in figure 6.
A printed circuit board was designed that would facilitate the
coupling of rows to form X signals and the coupling of columns to form
Y signals (see figure 7). The board is a double sided printed circuit
board incorporating plated through holes. The wire leads of the detectors
were soldered to the printed circuit board. A layer of teflon was


COUNTS/CHANNEL COUNTS/CHANNEL
43
MAUDERLI Walter
429- 72-4216
Detector : CdTe
Isotope: Co-57
ORTEC Pre-Amp.
Mod. 142 A
(followed by Ortec
Research Amp
Mod. 450 and
Tracor Northern
Multichannel Analyzer
Mod.TN-1705
1024 Ch .)
SPECIALLY DE
SIGNED Pre-Amp.
(followed by the same
Instrumentsas above)
WM 012078
Figure 12 Amplifier Spectrum Comparisons


COUNTS/CHANNEL COUNTS/CHANNEL
27
Detectors: CdTe
Isotope: Co-57
Bias: 100V
Preamplifier: Ortec 142A
Amplifier: Ortec 450(1.5^sec RCs)
Original Crystal Prior
To Cutting(ND035-6-3)
Leakage Current: 25nAs65V
One Of Small Crystals
After Cutting(ND035-6-3-A)
Leakage Current: 6nA965V
ENERGY I
122 KeV
Figure 4
Spectometry Performance Before And After Cutting


62
center column does show some activity since the edges of both lines
overlap into the center column. Figure 20 illustrates that two line
sources of different activities can be imaged although the distance
between the lines (2.5 mm) is less than the physical dimension of one
collimating tube if the lines are oriented over two adjacent columns.
Degradation of image quality is observed by the introduction of 0.43
inch and 0.93 inch of Plexiglass scattering media between the lines
and the detector matrix is illustrated in figure 21 and figure 22. As
discussed subsequently, scattered radiation was not accurately discriminated
against. Geometry also contributes to the image degradation.
It was interesting to note during use of parallel line sources,
that air bubbles created within the device when the lines were loaded
could be diagnosed by the camera as shown in figure 23.
The sensitivity of the camera is qualitatively illustrated in
figure 24. The image was acquired in a counting time of approximately
3.8 seconds.
To illustrate the ability of the camera to image volume sources,
a thyroid phantom was imaged. The thyroid image represented in figure
25 is actually the composite of 36 images. Each image was normalized to
a common time frame by correcting for counting time and for decay of
the radioactivity. The 5 mm cold spot in the upper left hand corner is
somewhat spread out. This occurred because the cold spot was located at
the juncture of four composite images. Subsequent images of the cold
spot, with the cold spot centered over the detector matrix produced a
much sharper and better defined image.
The response of the camera was checked for paralysis by utilizing
a variety of uniform plane source activities (to a maximum of 0.146


21
summation of two or more detection events are not necessary
in this camera since detection rates will be maintained at a
low level.
The advantage is further illustrated by an examination
of applicable attenuation coefficients. For 150 keV photons,
the attenuation in tin due to scattering (including coherent)
is approximately 25% of the total attenuation coefficient
(Hubbell, 1969). Values for tin are used, since the average
atomic number of CdTe and tin are the same. One of four
photons interacting in CdTe will undergo scattering. If the
CdTe detectors are not located in hole, many of the scattered
photons could be detected in adjacent detectors and hence
produce erroneous positional information. The reduction of
radiation scatter between detectors by the in hole detector
principle, is particularly important in this application since
the small size of the detector and the size and cost limitations
on amplifiers may make precise pulse height discrimination
against scattered radiation very difficult.
The fabrication of the camera is not greatly affected by
the in hole detector geometry since the detectors must have
electrical leads applied, must be separated by insulating media
and have regular overall physical dimensions for ease of
construction. It has been demonstrated that separation of
multiple Nal(Tl) detectors by absorbing septa reduces scatter
and produces finer resolution (Beck, 1975).
2) As shown in figure 2, for a point source located on the
collimator hole axis, the detection of primary or full energy


Figure 10 Analog Circuitry


PACKAGE
DESICIMAT10H
saimn
A.I.C.D.E.F.Y.HH CO 4098
Dual om shot
G.H.U.V
CD 4072
Dual 4 Input OR gat*
I,J,K,FF,GC
CD 4042
Quad docked D latch
L,M,AA,B8
CD 4030
Hex buffer
i,T
CD 4071
8 Input NOR gat*
O.P.Q.R
CD 4081
Quad 2 input AMD gate
S
CD 4011
Quad 2 Input HAND gat*
w,x
CD 4082
Dual 4 Input AMD gat*
CC.DO
3341
84 word-4 bit first in first
out serial eeaory(FIFO)
EE
m 7401
Quad 2 input HAND gat*
1M3084
Dlod*
Figure 14
Dig
ital Circuitry
(Page 1 of 2)


20
operated in a row, the errors associated with noise are
propagated and the FWHM is increased. For existing Anger
cameras the FWHM is on the order of 14 to 20 keV for 140 keV
photons. Since the average energy loss in one incoherent
scatter by a 140 keV photon is only 24 keV and occurs at a
photon scatter angle of approximately 75, there exists a
significant fraction of photon scatters that result in
relatively small energy loss with significant scatter angles.
The most probable photon scatters depart very little energy
and have small scatter angles. Scatter angles of 90 are
the least probable (approximately 33% of maximum probability
and have scattered photon energies near 110 keV). As the
scatter angle increases from 90 to 180, the scatter probability
associated with the angle increases from approximately 33% to
approximately 55% of the maximum probability.
For photon scatter events that occur in one detector,
adjacent detectors occupy 20% to 50% of the solid angle that a
scattered photon might enter. Since pulse height discrimination
cannot totally delete contributions from scattered photons,
individual CdTe detectors can be placed within the holes of
the collimator and hence the detectors can be separated by
collimator septa. The septa can then significantly attenuate
scatter occurring on the collimator hole surfaces or within
the detectors themselves, that might otherwise be counted in
adjacent detectors and contribute spurious data to the image.
Discrimination against pulses that have amplitude in excess
of the photopeak pulse height resulting from the coincident


BIOGRAPHICAL SKETCH
Jerry David Allison was born on June 18, 1948, in Brevard, North
Carolina. He attended local primary and elementary schools and entered
North Carolina State University in 1966. He was graduated in 1970 with
a Bachelor of Science in nuclear engineering. Mr. Allison was employed
at Newport News Shipbuilding in Virginia as a Radiological Control Engineer
from 1970 until 1975. In 1974, he earned the Master of Engineering
degree in mechanical engineering from Old Dominion University in Norfolk.
In 1975, he entered the University of Florida for graduate work in
medical radiation physics.
Mr. Allison is a licensed Professional Engineer in the Commonwealth
of Virginia and is certified in health physics by the American Board
of Health Physics.
He is married to the former Mary Jacqueline Andrews, also of
Brevard. The Allisons have two sons.
81


19
The advantage of CdTe is demonstrated by a comparison
of photopeak energy resolutions. The full width at half
maximum (FWHM) energy resolution for a 122 keV photopeak on
Anger cameras is on the order of 14 to 20 keV (Allemand, 1977).
The FWHM energy resolution of CdTe for 122 keV photons has
been reported to be as low as 3.8 keV (Siffert, 1975). For
the CdTe available for this research, the FWHM is on the order
of 8 to 10 keV for 140 keV photons.
4) In addition to the potential for improvement in pulse height
resolution described above, CdTe has a higher density and
higher atomic numbers resulting in increased photon attenua
tion per unit of detector thickness.
Advantages of In Hole Detectors
The advantages of locating the detectors within the collimator holes
rather than behind the collimator are as follows:
1) As discussed previously, discrimination against scattered
photons is desired. The discrimination is somewhat complicated
since full energy photon pulse heights vary statistically
about some mean value. Ideally, pulse height discrimination
would eliminate from consideration all scattered events due
to the reduced energy of the photons; however, there is a
statistical fluctuation around the true pulse heights that
complicates total discrimination against scattered photons.
For individual CdTe detectors the full width at half maximum
is on the order of 8 10 keV for 140 keV photons for the
detectors used in this research. When several detectors are


scatter radiation between matrix elements by the collimator septa that
separate detectors, enables improved resolution and imaging time.
The superior radiation detection characteristics of cadmium telluride
(in comparison to sodium iodide) could provide superior discrimination
against scattered radiation that contributes only spurious information
to the image. The availability of large numbers of small cadmium
telluride detectors that have uniform radiation detection properties would
make a clinical camera viable.
vi


29
Figure 6 CdTe Detector (With Leads Applied)


A B C D E F G
A-Base Plate
B-Spacer Block
C-Teflon Insulator
D-Printed Circuit Board
E-Teflon Insulator
F-Collimator
G-Cover Plate
Figure 9 Detector Head (Less Detectors)
36


64
Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: 0.43 inches of Plexiglas scattering media
IMPOSED BETWEEN COLLIMATOR AND SOURCE
Scale: arbitrary
Activities: lines have equal specific activities
XY CdTe Camera
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 21
Parallel Line Source Image (0.43 Inches
of Scattering Media)


16
several matrices where each matrix has its own electronics package.
Electronic storage devices are available that can temporarily store the
data as fast as they are produced. The size of a matrix is essentially
governed by the number of detectors that can be operated
without serious degradation of signal to noise ratio. Improvements
in detector properties would produce analogous increases in matrix size
possible.
Advantages of a CdTe Matrix
The application of CdTe as semiconductor radiation detectors in an
X Y matrix radioisotope camera eliminates some problems of systems in
corporating Nal(Tl) detectors. The advantages of CdTe in an X Y matrix
detector are summarized as follows:
1) "The elaborate assignment of X and Y coordinates in the
scintillation photomultiplier array method is completely
eliminated by the simple but ingeneous method of reading
these coordinates directly off the conduction surfaces of
the semiconductor itself, using the direct, spatially
quantitized digital output of the crystal" (Beck, 1975, p. 25).
The development of X Y positional information in the
Anger camera utilized analog signals to locate the center
of luminous intensity. As described in 2) below, the method
creates some inherent limitations in regards to determining
the actual position of radiation interaction in the Nal(Tl)
crystal. An X Y matrix of CdTe detectors can be used to
obtain a digital address (row number and column number where
signals were generated by a detection event) and the


50
lines. The output latches of FF and GG are buffered through buffer BB
to the inputs of first in first out serial memories (FIFO) CC and DD.
The FIFOs serve to assemble incoming data at one clock rate and transfer
the data out at a different clock rate. Each FIFO is a 64 word x 4 bit
storage device. Attempts to input data into the FIFO when all 64
registers are full are ignored. The FIFOs can easily be ganged to create
a composite FIFO of any dimension desired. When high, pin 1 of AND gate
W (after a delay through one shot AA for data set up) enables FIFOs CC
and DD to latch the data at their inputs. After data have been accepted
into the serial memory of the FIFOs, the FIFOs generate through NAND
gate EE a reset pulse that resets one shot Y to be ready for the next
data transfer.
Once data are entered in the FIFOs, data automatically "bubbles"
to the FIFO output. The microcomputer parallel input port (an eight
bit port) has two control signals that readily interface with the output
control lines of the FIFOs. The output ready lines of FIFOs CC and DD
activate NAND gate EE to create through buffer AA and one shot HH, a
ly sec pulse that pulls low the parallel data ready (PDR) line of the
microcomputer indicating that valid data are present at the port. The
microcomputer, under program control, inputs the data from the parallel
port and stores the data in memory for subsequent processing. Once
the data transfer into the microcomputer is complete, the parallel
input acknowledge line (PIAK) goes low to signal the FIFOs that the
data transfer is complete and that another data transfer may be initiated.
A master reset signal is activated by a switch, through one shot Y.
The FIFO memories are cleared and the FIFO control lines are initialized
when the master reset signal is activated.


63
mCi/cm^ of 99mTc). The counting rate was directly proportional to
source activity and no paralysis was evident.
The coincident time period was varied from near 0 to 5 yseconds
with no apparent change in image quality.
The background count rate was so low as to be inconsequential.
The X Y CdTe camera had substantial imaging capabilities as il
lustrated in the preceding figures. The results achieved as well as
difficulties encountered in achieving those results are discussed in
the next chapter.


6/17/08 9:2.1 AM
UF Librangs: Digital Dissertation Project
S)C\r\M/ /-)
Pk i
D
Printer! nr Tvnerl Nafne nf tloWriffht Hnlder/Licensee
A
Personal information blurred
Hal&3
Please print, sign and return to:
Cathleen Martyniak
UF Dissertation Project
Preservation Department
University of Florida Libraries
P.O. Box 117007
Gainesville, FL 32611-7007
mhtmlmain:
Page 2 of 2


34
Figure 8
Collimator


14
Other high density compound semiconductors are currently avail
able and the expectations for other high Z semiconductor radiation
detectors have been evaluated (Armantrout, 1977). Aluminum antimonide,
indium phosphate, and zinc telluride have had relatively little work
accomplished and ultimate technological feasibility is yet to be deter
mined. Mercuric iodide and gallium arsenide have been fabricated and
used, but do not excell in comparison to CdTe.
The electrical signal output of the X Y matrix is developed as
illustrated in figure 1 for a 25 element matrix. Electrical signals
are developed directly via the semiconductor radiation detector principle
rather than the less direct electrical signal generation of scintillation
radiation detectors. The X Y matrix signal configuration reduces the
number of outputs for an n x n matrix to a minimum of 2 n. Outputs
numbering in excess of 2 n may be required since the summation of
capacities or leakage currents in any one column or row may limit the
number of detectors per output.
The theoretical limit of resolution for the camera approaches the
centerline to centerline distance between adjacent matrix elements. The
ultimate resolution of a camera of this type can be selected to the
extent that the size of the detector and the size of the collimating
holes can be selected. Of course, as the physical size of the matrix
elements is reduced, the number of matrix elements and the associated
electronics are increased proportionately.
The sensitivity of the camera ultimately depends upon the duration
of pulses produced by the detectors and associated electronics and upon
the number of matrix elements. Large cameras would be composed of


UF Libraries: Digital Dissertation Project
6/17/08 9:2.1 AM
Dear Dr. Jerry D. Allison,
The George A. Smathers Libraries at the University of Florida has initiated a project to retrospectively digitize and
make available on the Internet any dissertation written by a University of Florida doctoral candidate and accepted by
the University of Florida. It is our goal to make the documents fully text searchable and easily harvested by Internet
search engines, allowing the full breadth and scope of scholarship produced at the University of Florida to be made
available across the world quickly and easily. The Library is bearing the full cost of the project.
We would like to add your dissertation, The application of cadmium telluride as in hole semiconductor radiation
detectors in an X Y matrix radioisotope camera /, published in 1978, to the project. In order to do so we need a
signed, legal original Internet Distribution Consent Agreement for our files. If you want your dissertation included in
the project, please print the Consent Agreement on the page below, sign it and mail it back to the Libraries at the
address listed.
We will keep you informed of the progress of your dissertation as it works its way through the project.
If you have any questions, please reply to UFdissertations@uflib.ufl.edu.
Thank you,
Cathy Martyniak, Project Coordinator
Christy Shorey, Project Technician
fin UNIVERSITY of
Ur FLORIDA
The Foundation for The Gator Nation
Internet Distribution Consent Agreement
In reference to the following dissertation:
AUTHOR: Allison, Jerry
TITLE: The application of cadmium telluride as in hole semiconductor radiation
detectors in an X Y matrix radioisotope camera / (record number: 74791)
PUBLICATION DATE: 1978
I, 1^as copyright holder for the aforementioned dissertation, hereby grant
specifte'ap'd limited aryhiVe and distribution rights to the Board of Trustees of the University of Florida and its agents. I authorize
the Ifnjtfersity of Florida to digitize and distribute the dissertation described above for nonprofit, educational purposes via the
Internet or successive technologies.
This is a non-exclusive grant of permissions for specific off-line and on-line uses for an indefinite term. Off-line uses shall be
limited to those specifically allowed by "Fair Use" as prescribed by the terms of United States copyright legislation (cf, Title 17,
U.S. Code) as well as to the maintenance and preservation of a digital archive copy. Digitization allows the University of Florida
to generate image- and text-based versions as appropriate and to provide and enhance access using search software.
This gp
igitized versions for commercial use or profit.
Imain:
Page 1 of 2


10
rather inefficient transfer of their scintillations to relatively large
photomultiplier tubes via fiber optic material.
The direct electrical signal generation of semiconductors is an
important contrast to the scintillation detection process of Nal(Tl)
detectors. In semiconductor detectors, the electron hole pairs created
via ionization are collected on electrodes due to the bias voltage
applied to the detector. In Nal(Tl) detectors, the scintillations
produced by excitation and ionization must pass through the detector,
be transferred through the light pipe to the photoelectric material of
the photomultiplier tube where the electrical signal is developed.
Cadmium telluride is a II-VI compound semiconductor having the
sphalerite crystal structure of zincblende. Cadmium is atomic number
48. Tellurium is atomic number 52. The compound, CdTe, has an average
atomic number of 50 and a specific gravity of 5.20. The electrical
property of CdTe that makes it particularly useful as a semiconductor
radiation detector is its energy bandgap at 300 degrees kelvin (K) as
compared to other semiconductor radiation detectors (1.5 eV for CdTe,
1.08 eV for silicon (SI), and 0.67 eV for germanium (GE)) (Palms, 1971).
The relatively large bandgap allows for operation of the detector at
room temperature without excessive thermal generation of charge carriers.
Excessive thermal generation of charge carriers results in a high
leakage current and poor signal to noise ratio which degrades detector
resolution (as occurs with Si and particularly with Ge when operated at
300K).
The physical properties of CdTe that make it particularly useful
are its relatively high atomic number (Z) (CdTe:50, Si:14, Ge:32) and


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
\
7
Ci\:>
ICCs- J rCl'
Walter Mauderli, Chairman
Professor of Nuclear Engineering
Sciences
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
W. Emmett Bolch
Professor of Nuclear Engineering
Sciences
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
Valerie A Brookeman
Associate Professor of Radiology
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
!ojx;
LV
/ /
(A.-l-
Lawrence T. Fitzgerald
Assistant Professor'of Nuclear
Engineering Sciences


xml record header identifier oai:www.uflib.ufl.edu.ufdc:UF0008974500001datestamp 2009-02-09setSpec [UFDC_OAI_SET]metadata oai_dc:dc xmlns:oai_dc http:www.openarchives.orgOAI2.0oai_dc xmlns:dc http:purl.orgdcelements1.1 xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.openarchives.orgOAI2.0oai_dc.xsd dc:title The application of cadmium telluride as in hole semiconductor radiation detectors in an X Y matrix radioisotope cameradc:creator Allison, Jerry Daviddc:publisher Jerry David Allisondc:date 1978dc:type Bookdc:identifier http://www.uflib.ufl.edu/ufdc/?b=UF00089745&v=00001000074791 (alephbibnum)04718263 (oclc)dc:source University of Floridadc:language English


CHAPTER 6
DISCUSSION OF RESULTS
The X Y CdTe camera concept offers the advantage of direct digital
localization of detected events, improvement of pulse height resolution
and attenuation of scatter within the detecting medium. As shown in the
previous chapter, the X Y CdTe camera produced images superior to other
existing devices. The images indicate that line source detection can
essentially be limited to the detectors directly beneath the line and
that imaging time is shorter than the other cameras tested. The per
formance of the X Y CdTe camera, however, was not without limitations.
The performance of the camera was limited by two important problems
1) The photopeak pulse heights of the individual CdTe detectors
varied by a factor of approximately two. The detector rows
were grouped according to pulse heights (i.e., each row had
detectors with similar pulse heights); however, as a result,
each detector column had a spectrum of photopeak pulse heights
ranging from very low to very high. Pulse height discrimina
tion for all columns was established at a level low enough
to accommodate detection of events in the detectors with very
small pulse heights. As a result, the detectors having
larger pulse heights produced a lot more counts than detectors
having small pulse heights.
2) Although at one time there were 25 detectors that exhibited
photopeaks, after a relatively short usage time, several of
69


41
This research utilized a 5 x 5 matrix of CdTe detectors and hence
required 10 amplifiers. The 10 amplifiers were fabricated onto one
printed circuit board and hence it was necessary to pay particular
attention to guarding the inputs of the amplifiers to minimize feedback
caused by voltage and current fluctuations in adjacent amplifiers.
A state of the art commercial preamplifier and amplifier were procured
for use in testing the CdTe detectors. The commercial amplifiers offer
low noise and large signal to noise ratio and the ability to control the
integration and differentiation of the signal. When using the commercial
amplifiers, the photopeak energy resolution is largely determined by the
noise generated within the detector. The commercial amplifiers were
used as a reference device for testing the detector properties and for
comparison of the inexpensive amplifiers developed for use with the
matrix.
As shown in figure 11 and figure 12, the low cost charge sensitive
amplifiers developed in this research offer signal to noise ratio and
photopeak resolution comparable to the state of the art commercial
amp!ifier.
Analog to Digital Circuitry
The analog to digital converter was designed to produce a digital
output signal whenever an input signal exceeds some reference level and
hence indicate that a detection event occurred. The reference level can
be adjusted to represent the lower limit or threshold of a photopeak and
hence pulse height discrimination can be performed. The digital outputs
of the analog to digital converter circuits are inputs to the digital


BIBLIOGRAPHY
Allemand R, Bouteiller P, Laval M: Present limitations of CdTe
detectors in nuclear medicine. Rev Phys Appl 12:365-367, 1977
Anger HO: Radioisotope cameras. Instrumentation in Nuclear Medicine.
New York, Academic Press, 1967, pp 485-552
Anger HO: Instruments: specific devices. Nuclear Medicine. New York,
HP Publishing Co, 1975, pp 29-40
Arbel FA: The second stage noise contribution of a nuclear pulse
amplifier. IEEE Trans Nucl Sci 15:2-5, 1968
Armantrout GA, Swierkowski SP,. Sherohman JW, Yee JH: What can be
expected from high z semiconductor detectors? IEEE Trans Nucl Sci
24:121-125, 1977
Aven M, Prener JS, Editors: Physics and Chemistry of 11-VI Compounds.
Amsterdam, North Holland Publishing Co, 1967
Beck RN: Instruments: basic principles. Nuclear Medicine. New York,
HP Publishing Co, 1975, pp 17-28
Causer DA: The design of parallel hole gamma camera collimators. Int J
Appl Radi at Iso 26:355-362, 1975
Detko JF: A prototype ultra pure germanium orthogonal strip gamma
camera. Med Radioisotopes Scintigraphy 1:241-254, 1972
El ad E: FET preamplifiers for semiconductor radiation detectors.
ISPRA Nuclear Electronics Symposium Proceedings, 1969 pp 21-34
Entine G, Garcia DA, Tow DE: Review of CdTe medical applications. Rev
Phys Appl 12:355-359, 1977
ESPI Catalog 690. Los Angeles, Electronic Space Products Incorporated
1969, p 181
Hoffer PB, Beck RN, Gottschalk A, Editors: Semiconductor Detectors in the
Future of Nuclear Medicine. New York, The Society of Nuclear
Medicine, 1971, p 287
Hubbell JH: Photon Cross Sections, Attenuation Coefficients, and Energy
Absorption Coefficient From 10 keV to 100 GeV. Washington,
National Bureau of Standards, 1969, NSRDS-NBS29
79


2
A unique new radioisotope camera that utilizes an X Y matrix of
cadmium telluride (CdTe) semiconductors as the detecting medium and
that has the detectors located within the collimating holes rather than
behind the collimator has been developed. The new camera offers im
proved spatial resolution and reduced imaging time in comparison to the
Anger camera.
The objectives of the research described herein were as follows:
1) To develop a radioisotope camera which utilizes a small
matrix of CdTe detectors located within collimator holes.
The camera design should lend itself to use in devices
utilizing a larger matrix.
2) To describe the performance of the camera and to compare
the performance to that of other applicable cameras,
particularly the Anger camera.
3) To estimate the feasibility of larger cameras that
would be of practical clinical significance.
The design, construction and performance of the new camera is
documented in subsequent chapters. The ultimate clinical feasibility of
a larger camera incorporating the same design principles is discussed.


35
dimension tolerance and hence it was necessary to hand select tubes
and heat press them into the block. The component parts of the detecting
head (less detectors) are as shown in figure 9. A sleeve of insulating
tubing was placed around each detector when inserted into the collimator
so as to electrically insulate the detector from the tantalum tubes.
An aluminum foil was placed over the top of the collimator so as to
maintain the detectors in a dark environment (operation in light fields
increases the detector leakage currents). Additional layers of lead were
added to the exterior of the collimator to provide additional shielding.
Analog Circuitry
The analog portion of the electronics (i.e., that portion that
collects the charge from the rows and columns connecting the top electrodes
and bottom electrodes, respectively, and converts the charge to a voltage
representative of the energy deposited in a detection event) was
designed to meet the following criteria:
1) To produce a voltage pulse that is proportional in amplitude
to the charge pulse at the amplifier input and hence to the
energy deposited in a detection event.
2) To minimize the number and cost of components so as to
facilitate the hybridization that would be required for
clinically significant devices.
It is desirable to minimize the number of components since an n x n
matrix would require a minimum of 2n amplifiers. A clinical device
would require a large number of amplifiers and therefore in order to
minimize cost and to facilitate design hybridization, it is desirable to
use the fewest components commensurate with the application.


73
In summary, the X Y CdTe camera developed excellent resolution
and sensitivity. The direct digital localization was primarily
responsible for the success; however, in the absence of pulse height
resolution capabilities, the attenuation of scatter between elements in
the detector matrix became more important.


CHAPTER 4
CAMERA DESIGN AND CONSTRUCTION
An X Y matrix radioisotope camera was designed and constructed in
accordance with objectives set out in previous chapters. A block
diagram of the resultant camera is presented in figure 3. In the fol
lowing sections, each component of the camera will be discussed in
some detail, with emphasis upon the objectives for the design and con
struction of each component.
Detector Matrix
The detectors utilized in the matrix were cubes of approximately
2 mm side length. The detectors were cut from large crystals
(2 mm x 2 mm x 7 mm) using a wire saw (0.003 diameter stainless steel
wire impregnated with 8 micron size diamonds). The crystal surfaces
were not treated (lapped or polished) after cutting (some of the
large crystal surfaces were not treated either). The cubes had
platinum electrodes applied to an area on two parallel opposed sides
(the electrodes did not extend to the edges of the crystals on the
large crystals or on the small cubes). The large detectors that were
cut to produce the small cubes were selected for the lowest noise,
best photon spectra and best physical characteristics (i.e., a
regular shape with electrode surfaces approaching crystal edges)
from approximately 130 available.
24


77
8) High speed TTL FIFOs are currently available that input
clock rate capabilities in excess of 10 MHZ. By ganging
large numbers of these relatively inexpensive memories, it
would be possible to store data as fast as the detector
matrix produces it and hence maximize camera sensitivity.
The data could later be transferred into computer memory at
the much slower rate required for that data transfer.
9) The image might be improved by inserting interpolated data
between actual detector measured data, based on the average
of measured data from adjacent elements. This could tend to
smooth out the transition of data between measured data
elements. An alternate gray level calculation scheme could
also be used. Rather than simply identifying sixteen levels
between zero and the maximum count obtained, it is possible
to define sixteen gray levels between the minimum count and
maximum count or base the image around an average count.
These procedures will not alter the basic data upon which the
image is based, but may produce an image that is aesthetically
more pleasing or easier to visualize. It should also be noted
that color images can also be produced.
The basic clinical feasibility of a camera based on the
concept of constructing a large matrix of small detectors depends
upon the availability of large numbers of small detectors that
have essentially the same radiation detection properties. The
variability that exists, even in specially selected or grouped
CdTe detectors, is too large to fabricate a CdTe matrix camera


THE APPLICATION OF CADMIUM TELLURIDE AS IN HOLE SEMICONDUCTOR
RADIATION DETECTORS IN AN X Y MATRIX RADIOISOTOPE CAMERA
BY -
JERRY DAVID ALLISON
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE
UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1978

To My Parents
a£

ACKNOWLEDGMENTS
The members of my advisory committee (Drs. W. Emmett Bolch, Valerie
A. Brookeman, Lawrence T. Fitzgerald, Walter Mauderli, Genevieve S.
Roessler and Henri A. Van Rinsvelt) were instrumental in the formula
tion, accomplishment and documentation of this research. Dr. Mauderli
served as chairman of the committee and provided special assistance
with analog electronics. Dr. Fitzgerald provided special assistance
with digital electronics and microcomputer applications. Dr. Bolch,
through the Department of Environmental Engineering Sciences, provided
a graduate assistantship during my first year at the University.
Dr. Genevieve Roessler counseled me with regard to course work and
helped me find appropriate assistantships throughout my graduate study
at the University.
Dr. Clyde M. Williams, as Chairman of the Department of Radiology,
provided a graduate assistantship for myself and financial support for
my research project.
The person most responsible for making this dissertation possible
is my wife, Jacquie. Jacquie and I left rather secure professional
careers and a very pleasant social environment in Newport News,
Virginia, so that I could pursue a doctorate at the University of Florida.
Not only did Jacquie provide moral support but also supported our family
financially by teaching in the secondary schools of Alachua County.
She bore our second child and completed her master's degree during our
three year stay in Gainesville. Jacquie made this dissertation possible.
iii

TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS Hi
ABSTRACT v
CHAPTER
1. INTRODUCTION 1
2. HISTORICAL DEVELOPMENT 3
3. CAMERA PROPERTIES 7
Advantages of a CdTe Matrix 16
Advantages of In Hole Detectors 19
Disadvantages 23
4. CAMERA DESIGN AND CONSTRUCTION 24
Detector Matrix 24
Collimator 31
Analog Circuitry 35
Analog to Digital Circuitry- 41
Digital Circuitry 46
Microcomputer 51
Videographics Interface 53
Summary 54
5. RESULTS 55
6. DISCUSSION OF RESULTS 69
7. CLINICAL FEASIBILITY 74
BIBLIOGRAPHY 79
BIOGRAPHICAL SKETCH
81

Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
THE APPLICATION OF CADMIUM TELLURIDE AS IN HOLE SEMICONDUCTOR
RADIATION DETECTORS IN AN X Y MATRIX RADIOISOTOPE CAMERA
By
Jerry David Allison
August 1978
Chairman: Walter Mauderli, D. Sc.
Major Department: Nuclear Engineering Sciences
A radioisotope camera that has superior performance in comparison
to conventional devices, has been designed, constructed, and tested.
Radioisotope cameras are instruments used for imaging the distribution
of radioisotopes in vivo. A camera creates a two dimensional image of
the three dimensional radioisotope distribution within an object. Con
ventional radioisotope cameras are mainly of the Anger type. The radia
tion detector in an Anger camera is a sodium iodide scintillation crystal
that is located in back of a collimator. The collimator creates a
spatial relationship between the image and the radioisotope distribution.
The new camera is a small experimental camera that has a matrix of
cadmium telluride semiconductors as the detecting medium. The detectors
are located within the collimating holes (rather than behind them).
Coincident outputs from each row and column are used to digitally
identify the individual detector in which the energy deposition occurred.
Direct digital localization of detected events and the attenuation of
v

scatter radiation between matrix elements by the collimator septa that
separate detectors, enables improved resolution and imaging time.
The superior radiation detection characteristics of cadmium telluride
(in comparison to sodium iodide) could provide superior discrimination
against scattered radiation that contributes only spurious information
to the image. The availability of large numbers of small cadmium
telluride detectors that have uniform radiation detection properties would
make a clinical camera viable.
vi

CHAPTER 1
INTRODUCTION
Radioactive tracer studies are an integral part of modern medical
diagnostic procedures. By administering radioactive tracers (radio
pharmaceuticals) that are specific for some physiological function,
physicians are able to determine the level of that function within a
patient in terms of physical and chemical phenomena and hence determine
the site and rate of important biological processes. Radioisotope
cameras are the instruments that enable many of the relatively nonin-
vasive radioactive tracer studies that are currently prevalent in nuclear
medicine. It is because of the fact that improvements in spatial or
temporal resolution of radioisotope cameras have the potential of making
better diagnostic information available to the physician as well as
reducing radiation dose to the patient, that the research described
herein was undertaken and is important.
Radioisotope cameras image the distribution of radioisotopes in
vivo. A camera creates a two dimensional image of the three dimensional
radioisotope distribution within an object. Radioisotope cameras are
mainly of the gamma ray scintillation type (Anger camera). Anger
cameras employ a sodium iodide [Nal(Tl)] scintillation crystal of 1/8
to 1 inch thickness as the detecting medium. The Nal(Tl) crystal is
located in back of a collimator which creates a spatial relationship
between the image and the object.
1

2
A unique new radioisotope camera that utilizes an X Y matrix of
cadmium telluride (CdTe) semiconductors as the detecting medium and
that has the detectors located within the collimating holes rather than
behind the collimator has been developed. The new camera offers im
proved spatial resolution and reduced imaging time in comparison to the
Anger camera.
The objectives of the research described herein were as follows:
1) To develop a radioisotope camera which utilizes a small
matrix of CdTe detectors located within collimator holes.
The camera design should lend itself to use in devices
utilizing a larger matrix.
2) To describe the performance of the camera and to compare
the performance to that of other applicable cameras,
particularly the Anger camera.
3) To estimate the feasibility of larger cameras that
would be of practical clinical significance.
The design, construction and performance of the new camera is
documented in subsequent chapters. The ultimate clinical feasibility of
a larger camera incorporating the same design principles is discussed.

CHAPTER 2
HISTORICAL DEVELOPMENT
The radioisotope camera developed in this research is a unique
new design that has no direct predecessors. Some of the concepts utilized
in the camera have been tried in other types of radioisotope cameras;
however, the camera developed in this research is the first to employ
CdTe semiconductor detectors in anXY matrix and the first to locate
the detectors within the collimating holes. The research was original
in that a review of pertinent literature revealed the following:
1) Cadmium telluride detectors have not previously been
employed in an X Y matrix gamma camera radioisotope
imaging system. The idea to utilize CdTe in a radio
isotope imaging device is not original to this research.
Work is under way at several laboratories in regard to
the development of radioisotope cameras utilizing semi
conductor detectors (Entine, 1977). The use of CdTe
detectors in an X Y matrix, thus reducing the number of
outputs, has been offered for consideration (Allemand,
1977). One rudimentary imaging system utilizing an array
of CdTe detectors behind a parallel hole collimator was
developed (Zanio, 1977). The system utilized (n) outputs
(for an n x n matrix) and a nonuniform electric field
geometry. The detectors were joined electrically in an
attempt to fabricate a large volume detector. Although
the system functioned, it was not designed as a radioisotope
3

4
camera and is not really applicable to this research.
The application of CdTe to radioisotope cameras is
sufficiently diversified to assure that each study is
unique and significant.
2) Semiconductor detectors have not been employed in an inhole
matrix geometry. Cameras have been made that utilize
orthogonal strip germanium detectors (Detko, 1972). The
cameras have demonstrated significant resolution but are
somewhat limited in size. Fabrication of larger cameras from
several orthogonal strip detectors is underway. The
germanium requires cooling during operation and since each
orthogonal strip detector is actually one detector with many
leads, some cross talk between elements exists (particularly
when adjacent conducting strips are in close proximity for good
resolution). The orthogonal strip arrangement does not lend
itself to shielding between elements so as to reduce scattered
radiation contributions. The extension of the orthogonal
conducting strips beyond the ideal sensitive volume of the
detectors creates substantial nonuniform electric fields near
the edges of each detector. Within the regions of nonuniform
electric fields, the field strengths are less than in the
more sensitive regions. As a result, incomplete charge
collection occurs and substantial numbers of pulses with
reduced pulse height are produced that must be discriminated
against. The limitation on spacing between adjacent con
ducting strips creates substantial gaps between detectors

5
that reduce the sensitive area of the matrix (in comparison
to total crystal area). Orthogonal strip cameras constructed
to date utilize large resistor divider networks to produce
one analog signal representing X position and one analog
signal representing Y position, rather than direct digital
localization of the detector in which a detection event
occurred.
3) Gamma cameras employing multiple Nal(Tl) crystals have not
been substantially successful in comparison to the Anger
camera. One such device, the autofluoroscope, was composed
of 294 Nal(Tl) crystals, each of which was coupled to an X
position and Y position photomultiplier tube (PMT) by long
light pipes. The autofluoroscope could operate at high count
rates but had poor photopeak energy resolution caused by
inefficient transfer of light through the light pipes (Beck,
1975). The poor energy resolution reduced the ability to
discriminate against scattered photons.
A camera, emoloying 254 Nal(Tl) crystals, was developed
recently (Sveinsdottir, 1977). The spatial resolution was
essentially the same as previous multidetector scintillation
cameras but the temporal resolution was substantially improved
by utilizing parallel event processing in lieu of sequential
event processing. The number of crystals used remains some
what limited by the use of one light pipe, one PMT, amplifier,
discriminator, buffer, etc. per crystal.
Another such device was the Quantascope camera which
employed 2515 cesium iodide CsI(Tl) crystals that were

6
separated by lead strips in order to reduce the detection
of scattered radiation. The quantascope utilized an
image intensifier rather than PMTs. The camera produced
poorer contrast than is available with an Anger camera and
required matched crystal selection (Beck, 1975).
4) Another CdTe radioisotope camera was recently constructed;
however, the application involves computer-assisted recon
struction techniques for the development of a two dimensional
representation of the three dimensional radioisotope distri
bution (Luthmann, 1978). Data is obtained from the n detector
rows situated within a parallel plate collimator which is
rotated about its central axis.
Most of the radioisotope cameras that have been developed, offer
improvement in some parameter as compared to the Anger camera. A com
plete evaluation of the devices reveals that a multipurpose device that
can offer substantial improvement in sensitivity, spatial and temporal
resolution and hence an improvement in diagnostic information obtained
or a reduction in patient exposure, has not been developed.

CHAPTER 3
CAMERA PROPERTIES
Direct digital localization of detected events in the X Y CdTe
camera developed in this research is an advantage in comparison to the
Anger camera. Also, the utilization of semiconductor radiation detectors
(particularly CdTe) and the placement of the detectors within collimating
holes offer some advantages over conventional radioisotope cameras.
To better appreciate the advantages, some of the properties of the
Anger camera and the X Y CdTe camera will be examined.
The deposition of energy, by radiation, within the Nal(Tl) crystal
of an Anger camera results in scintillations proportional to the ioniza
tion and excitation produced. The scintillations are detected by an
array of PMTs which convert the light into electrical signals proportional
in amplitude to the quantity of light incident upon the PMTs. The
location of the energy deposition is determined via a capacitive net
work that divides the pulse amplitudes according to the distribution of
luminescence with respect to the positions of the PMTs. The resultant
pulses are then converted to visible light flashes at appropriate sites
on a cathode ray tube (CRT). The position of the flashes on the CRT
are determined by difference circuits or ratio circuits.
For purposes of forming an image of the distribution of radio
activity within an object, only photons created within the object that
reach the sensitive volume of the detector without scattering (coherent
7

8
and incoherent) are useful (excluding the positron camera principle).
The origin of scattered photons detected by a camera cannot be limited
to a solid angle as is possible with unscattered photons. In order to
eliminate some of the signals generated by photons that have been in
coherently scattered (and hence have less energy), a pulse height
analyzer is incorporated in conventional cameras. The pulse height
analyzer discriminates against those signals whose amplitude is not
large enough to indicate that a full energy photon (i.e., a monoenergetic
photon whose energy has not been reduced by scatter) was absorbed by
the crystal. In practice, it is not possible to precisely discriminate
against scattered photons since even full energy photon pulse heights
exhibit statistical variation about some mean value. The signal caused
by a scattered photon or the coincident summation signal caused by
several scattered photons may have an amplitude within the statistical
variation of full energy photon signals. It is possible to discriminate
against some of the scattered photons by setting a threshold pulse
height that is just below the photopeak of the full energy photons.
Coherently scattered photons are essentially full energy photons that
are scattered into some solid angle without significant energy transfer
occurring. Coherently scattered photons do not contribute useful
informaton to the image formation process but cannot be discriminated
against. A collimator is incorporated in the Anger camera to create a
spatial relationship between the image created by full energy absorp
tion events in the crystal and the distribution of radioactivity in the
object.
"Semiconductor detectors offer many advantages for clinical and
experimental uses, chiefly because of their small size and excellent

9
energy resolution. The energy resolution of a solid state detector
is about twenty times higher than that of a conventional Nal(Tl)
scintillation detector. In addition, the elimination of scattered
radiation contributes to the improvement of image quality. The semi
conductors are considerably more efficient than Geiger probes and their
bias voltage is lower than that required by any other probe.
At present, solid state detectors are being used in the imaging
techniques; however, thus far attempts to make a high resolution semi
conductor gamma camera met with considerable difficulties" (Meyer, 1975,
p. 697).
Gas ionization chambers can also be fabricated into relatively
small dimensions but are limited in usefulness by the relatively inef
ficient deposition of energy to the gas contained therein.
Attempts at utilizing semiconductor detectors in high resolution
gamma cameras have consisted thus far of placing the detector close to
the back surface of the collimator. The research described herein,
utilizes the semiconductor detectors within the holes of the collimator
device rather than in back of the collimator device.
It is important that the collimator be located as close as possible
in order to prevent photons scattered on septa surfaces from being
detected in adjacent detectors and producing counts that do not give
an accurate representation of the origin of the photon.
The in hole configuration utilizes the small size into which semi
conductor detectors can be fabricated. Nal(Tl) detectors can be
fabricated into relatively small detectors but their use is somewhat
limited by the requirement that they be hermetically sealed and the

10
rather inefficient transfer of their scintillations to relatively large
photomultiplier tubes via fiber optic material.
The direct electrical signal generation of semiconductors is an
important contrast to the scintillation detection process of Nal(Tl)
detectors. In semiconductor detectors, the electron hole pairs created
via ionization are collected on electrodes due to the bias voltage
applied to the detector. In Nal(Tl) detectors, the scintillations
produced by excitation and ionization must pass through the detector,
be transferred through the light pipe to the photoelectric material of
the photomultiplier tube where the electrical signal is developed.
Cadmium telluride is a II-VI compound semiconductor having the
sphalerite crystal structure of zincblende. Cadmium is atomic number
48. Tellurium is atomic number 52. The compound, CdTe, has an average
atomic number of 50 and a specific gravity of 5.20. The electrical
property of CdTe that makes it particularly useful as a semiconductor
radiation detector is its energy bandgap at 300 degrees kelvin (K) as
compared to other semiconductor radiation detectors (1.5 eV for CdTe,
1.08 eV for silicon (SI), and 0.67 eV for germanium (GE)) (Palms, 1971).
The relatively large bandgap allows for operation of the detector at
room temperature without excessive thermal generation of charge carriers.
Excessive thermal generation of charge carriers results in a high
leakage current and poor signal to noise ratio which degrades detector
resolution (as occurs with Si and particularly with Ge when operated at
300K).
The physical properties of CdTe that make it particularly useful
are its relatively high atomic number (Z) (CdTe:50, Si:14, Ge:32) and

11
its relatively high specific gravity (CdTe:5.20, Si:2.4, Ge:5.35). The
relatively high Z and specific gravity enhance the energy absorption
efficiency of CdTe.
The CdTe detectors used in this research are p type, chlorine doped
for compensation and have chemically deposited platinum electrodes.
Platinum is frequently used as the electrode material since its work
function is relatively high which helps to prevent polarization phenomenon
(Siffert, 1976).
The electron and hole mobilities are on the order of 1000 and
2
80 cm /V-sec, respectively. The electron and hole mobi1ity-trapping
-3 -4 2
time products are on the order of 10 and 10 cm /V respectively. The
detectors are characterized as homogeneous semiconductor radiation
detectors. The crystallographic orientation of the crystals to be used
is not precisely known. The crystals grow preferentially in the [111]
direction. Detector slices are cut perpendicularly to the direction of
growth and electrodes are applied across what is probably the [111]
principal axis. "The effect of crystallographic orientation of the
detector was examined for the applied electric bias along three
principal directions--[111], [110] and [100]--but no significant dif
ferences were observed" (Siffert, 1976, p. 159).
The pulse shape produced by the detectors is a function of the
position and orientation of charges liberated. Ideally the charge
liberation would be parallel to the electrodes and very near the negative
electrode, resulting in rapid, simultaneous collection of all charges.
If the charge liberation is perpendicular to the electrodes, the charges
nearest the electrodes are collected first and so forth, such that

12
charge collection occurs over some time period. For charge liberation
near the negative electrode, high mobility elecrons traverse the
entire detector thickness and the low mobility holes traverse a much
shorter distance; hence charge collection requires less time than
those events in which holes must traverse longer distances. Ideally
the CdTe detectors should be irradiated with a normal incidence on the
negative electrode.
The approximate physical dimensions of the detectors available for
this research were 2x2x7 mm. The electrodes were two opposed sides
of 2 x 7 mm dimensions. The linear attenuation coefficient of CdTe
is approximately 0.36/mm for 140 keV photons and hence a 2 mm thick
detector should attenuate 60 percent of incident 140 keV photons (Hoffer,
1971). Photons of 140 keV are considered throughout this dissertation
since they are produced in the decay of technetium-99m (Tc-99m) which
is widely used in radioisotope imaging studies. The 7 mm detectors
will be cut into 3 detectors of approximate dimensions 2x2x2 mm.
The 60 percent attenuation of the photons of interest is sufficient
to demonstrate the characteristics of the X V CdTe camera.
The stability of the detector characteristics with time is important
since polarization phenomenon (i.e., changes in the internal electric
field with time) have frequently been found in CdTe detectors. It has
been reported that some p type, chlorine doped CdTe detectors with
platinum contacts are totally stable with time, do not exhibit polariza
tion and are sensitive throughout their volume (Serreze, 1974).
The bias voltage applied to the detectors will determine the
mean free path of the electrons and holes. To maximize charge collection,

13
the mean free path of the charge carriers should be less than the col
lection distance or detector width. For the best photopeak energy
resolution in a 2 mm detector thickness, the bias voltage applied should
be on the order of 500 volts. The detector generated noise due to
increased leakage current caused by electric fields as high as 2500 V/cm
make operation impractical at that level. It is recommended that the
bias voltage be maintained at the maximum practical voltage (Malm,
1973). For a practical bias voltage of only 500 V/cm, the charge col
lection time for a 2 mm detector would range from approximately 0.4 ysec
to 5 ysec. The decay time of a Nal(Tl) scintillator is approximately
0.3 ysec.
The detector thickness is essentially limited by the field strength.
Reasonable spectrometer performance for the CdTe available occurs
around 325 to 500 V/cm. For a minimum trapping time of 6 ysec and
2
mobilities of 80 and 1000 cm /V-sec for holes and electrons respectively,
the detector thickness is limited to about 2 mm. For higher field
strengths, noise degrades the spectrometer performance. Further
development of CdTe could enable higher spectrometer field strengths
and hence thicker detectors.
Cadmium telluride has many other properties which are not of great
interest in semiconductor radiation detector applications; however,
for completeness some of those properties are listed as follows:
piezoelectric, photovoltaic, magnetoresistive, pyroelectric, electro
optic (Aven, 1967). A listing of the main applications of CdTe is as
follows: infrared windows, electro-optic modulator, Gunn effect high
frequency current oscillator, piezoelectric devices, luminescent devices,
solar cells and nuclear detectors (Wald, 1977).

14
Other high density compound semiconductors are currently avail
able and the expectations for other high Z semiconductor radiation
detectors have been evaluated (Armantrout, 1977). Aluminum antimonide,
indium phosphate, and zinc telluride have had relatively little work
accomplished and ultimate technological feasibility is yet to be deter
mined. Mercuric iodide and gallium arsenide have been fabricated and
used, but do not excell in comparison to CdTe.
The electrical signal output of the X Y matrix is developed as
illustrated in figure 1 for a 25 element matrix. Electrical signals
are developed directly via the semiconductor radiation detector principle
rather than the less direct electrical signal generation of scintillation
radiation detectors. The X Y matrix signal configuration reduces the
number of outputs for an n x n matrix to a minimum of 2 n. Outputs
numbering in excess of 2 n may be required since the summation of
capacities or leakage currents in any one column or row may limit the
number of detectors per output.
The theoretical limit of resolution for the camera approaches the
centerline to centerline distance between adjacent matrix elements. The
ultimate resolution of a camera of this type can be selected to the
extent that the size of the detector and the size of the collimating
holes can be selected. Of course, as the physical size of the matrix
elements is reduced, the number of matrix elements and the associated
electronics are increased proportionately.
The sensitivity of the camera ultimately depends upon the duration
of pulses produced by the detectors and associated electronics and upon
the number of matrix elements. Large cameras would be composed of

Figure 1 CdTe Detector Matrix
Bbiththtii
KFcPcFa-
tHtHta-ob
0500a
SffiB
O - I
cn
Vbias (-) 1.5N

16
several matrices where each matrix has its own electronics package.
Electronic storage devices are available that can temporarily store the
data as fast as they are produced. The size of a matrix is essentially
governed by the number of detectors that can be operated
without serious degradation of signal to noise ratio. Improvements
in detector properties would produce analogous increases in matrix size
possible.
Advantages of a CdTe Matrix
The application of CdTe as semiconductor radiation detectors in an
X Y matrix radioisotope camera eliminates some problems of systems in
corporating Nal(Tl) detectors. The advantages of CdTe in an X Y matrix
detector are summarized as follows:
1) "The elaborate assignment of X and Y coordinates in the
scintillation photomultiplier array method is completely
eliminated by the simple but ingeneous method of reading
these coordinates directly off the conduction surfaces of
the semiconductor itself, using the direct, spatially
quantitized digital output of the crystal" (Beck, 1975, p. 25).
The development of X Y positional information in the
Anger camera utilized analog signals to locate the center
of luminous intensity. As described in 2) below, the method
creates some inherent limitations in regards to determining
the actual position of radiation interaction in the Nal(Tl)
crystal. An X Y matrix of CdTe detectors can be used to
obtain a digital address (row number and column number where
signals were generated by a detection event) and the

17
amplitude of the signals produced can be analyzed by
pulse height discrimination to eliminate some of the events
caused by scattered photons.
2) The X Y CdTe camera will not produce distortion near the
edges of the matrix. The generation of X Y positional informa
tion by the capacitive network pulse divider of the Anger
camera is based on the center of luminous intensity incident
on the PMT array. For detections that result in sufficient
luminescence production to be indicative of full energy
photons, events are displayed on the CRT in positional rela
tion to the effective center of luminescence incident on the
PMTs. Due to inhomogeneities and defects in the Nal(Tl)
crystal and due to reflection of light at the crystal
boundaries, the center of luminescence incident on the PMTs
is not in the exact position of energy deposition by the
radiation. This distortion is relatively small in the
center of the crystal but increases near crystal edges. The
distortion is as high as 30% near the edges of some Anger
cameras (Moretti, 1974). Photons that undergo multiple
incoherent scatters within the Nal(Tl) crystal can cause
significant differences in the location of the center of
luminous intensity and the point of the photon's initial
interaction (which contains the positional information).
For a matrix of CdTe detectors, the location of a
photon detection is determined to be within a particular
detector via the X and Y address produced. The positional
information will not be distorted near the edge of the matrix

18
since the output is an address developed by the detector in
which the energy is deposited. The detectors are separated
from each other by the collimator septa and hence the probability
that a photon can deposit energy in more than one detector
is substantially reduced.
3) Improved photopeak resolution is possible with CdTe. In
scintillation detectors, the deposition of energy in the crystal,
by radiation, results in ionization and excitation. The energy
is converted to light by the luminescence process. The light
travels through the crystal and by means of a light pipe and
reflectors is transmitted to the photocathode of the PMT. The
light is absorbed by the photocathode resulting in photo
electron emission. An electron multiplication process
produces a large pulse at the output of the PMT. Each step
in the process has its own efficiency and statistical varia
tion. The statistical variation of the PMT output pulse is
the propagation of the variations of the individual steps
involved.
The semiconductor radiation detection principle of CdTe
is somewhat simpler in that the ion pairs (created by energy
deposition) are collected at the electrodes due to the electric
field developed across the detector by the bias voltage
(assuming trapping and recombination are prevented). The con
version of ion pairs to light and light to electrons and the
associated statistical variations are eliminated. The charge
collected at the electrodes is amplified into a large pulse.

19
The advantage of CdTe is demonstrated by a comparison
of photopeak energy resolutions. The full width at half
maximum (FWHM) energy resolution for a 122 keV photopeak on
Anger cameras is on the order of 14 to 20 keV (Allemand, 1977).
The FWHM energy resolution of CdTe for 122 keV photons has
been reported to be as low as 3.8 keV (Siffert, 1975). For
the CdTe available for this research, the FWHM is on the order
of 8 to 10 keV for 140 keV photons.
4) In addition to the potential for improvement in pulse height
resolution described above, CdTe has a higher density and
higher atomic numbers resulting in increased photon attenua
tion per unit of detector thickness.
Advantages of In Hole Detectors
The advantages of locating the detectors within the collimator holes
rather than behind the collimator are as follows:
1) As discussed previously, discrimination against scattered
photons is desired. The discrimination is somewhat complicated
since full energy photon pulse heights vary statistically
about some mean value. Ideally, pulse height discrimination
would eliminate from consideration all scattered events due
to the reduced energy of the photons; however, there is a
statistical fluctuation around the true pulse heights that
complicates total discrimination against scattered photons.
For individual CdTe detectors the full width at half maximum
is on the order of 8 10 keV for 140 keV photons for the
detectors used in this research. When several detectors are

20
operated in a row, the errors associated with noise are
propagated and the FWHM is increased. For existing Anger
cameras the FWHM is on the order of 14 to 20 keV for 140 keV
photons. Since the average energy loss in one incoherent
scatter by a 140 keV photon is only 24 keV and occurs at a
photon scatter angle of approximately 75, there exists a
significant fraction of photon scatters that result in
relatively small energy loss with significant scatter angles.
The most probable photon scatters depart very little energy
and have small scatter angles. Scatter angles of 90 are
the least probable (approximately 33% of maximum probability
and have scattered photon energies near 110 keV). As the
scatter angle increases from 90 to 180, the scatter probability
associated with the angle increases from approximately 33% to
approximately 55% of the maximum probability.
For photon scatter events that occur in one detector,
adjacent detectors occupy 20% to 50% of the solid angle that a
scattered photon might enter. Since pulse height discrimination
cannot totally delete contributions from scattered photons,
individual CdTe detectors can be placed within the holes of
the collimator and hence the detectors can be separated by
collimator septa. The septa can then significantly attenuate
scatter occurring on the collimator hole surfaces or within
the detectors themselves, that might otherwise be counted in
adjacent detectors and contribute spurious data to the image.
Discrimination against pulses that have amplitude in excess
of the photopeak pulse height resulting from the coincident

21
summation of two or more detection events are not necessary
in this camera since detection rates will be maintained at a
low level.
The advantage is further illustrated by an examination
of applicable attenuation coefficients. For 150 keV photons,
the attenuation in tin due to scattering (including coherent)
is approximately 25% of the total attenuation coefficient
(Hubbell, 1969). Values for tin are used, since the average
atomic number of CdTe and tin are the same. One of four
photons interacting in CdTe will undergo scattering. If the
CdTe detectors are not located in hole, many of the scattered
photons could be detected in adjacent detectors and hence
produce erroneous positional information. The reduction of
radiation scatter between detectors by the in hole detector
principle, is particularly important in this application since
the small size of the detector and the size and cost limitations
on amplifiers may make precise pulse height discrimination
against scattered radiation very difficult.
The fabrication of the camera is not greatly affected by
the in hole detector geometry since the detectors must have
electrical leads applied, must be separated by insulating media
and have regular overall physical dimensions for ease of
construction. It has been demonstrated that separation of
multiple Nal(Tl) detectors by absorbing septa reduces scatter
and produces finer resolution (Beck, 1975).
2) As shown in figure 2, for a point source located on the
collimator hole axis, the detection of primary or full energy

22
MAUDERLI Woller
429-72-4216
2 POINT SOURCES
ANGER TYPE
CAMERA
LEAD HOLES
S : Scatter orig. in
detector
R : Scatter orig. in
collimator
TUNGSTEN
PLATES
DETECTOR
lOverlapping sector of primary photon beam
reducing resolution
-reducing resolution*
PARALLEL
PLATE TYPE
CAMERA
DETECTOR
None of the [
above Problems 1
* Pulse-height discrimination does not eliminate scatter radiation when
the scatter angle is less than the following values: In case of Tc-99m
Lower discriminator level 5% below photon energy: ~ 50
10% : ~ 70
20% :~I00
(for Compton
scatter)
WM 012578
Figure 2 Detector-Collimator Arrangements

23
photons is essentially restricted to within the physical
dimensions of the detector located in the collimator hole.
For a Nal(Tl) crystal located behind the collimator, a detec
tion could occur beneath an adjacent collimator hole.
Scattered photons (scattered on collimator surfaces or within
the detector) are significantly attenuated by the collimator
septa that separate individual detectors.
Pisadvantages
The application is not without some limitations. The disadvantages
are as follows:
1) A substantial gap will exist between adjacent detectors
resulting in a reduced detector area as compared to a solid
crystal such as an Anger camera employs. The gap produces a
sensitive detector area of approximately 50% of the overall
camera dimensions as compared to 65 to 80% for the Anger camera.
The sensitivity is therefore reduced somewhat (for equivalent
detector thicknesses). The ratio of the sensitive detector
area to the camera face area could be increased with more
sophisticated fabrication technology.
2) The theoretical limit of resolution is limited by centerline
to centerline spacing of the detectors. In this application,
the centerline to centerline distance may be in excess of 3 mm.
Should the application prove successful, the distance could be
reduced by smaller detectors and gaps.

CHAPTER 4
CAMERA DESIGN AND CONSTRUCTION
An X Y matrix radioisotope camera was designed and constructed in
accordance with objectives set out in previous chapters. A block
diagram of the resultant camera is presented in figure 3. In the fol
lowing sections, each component of the camera will be discussed in
some detail, with emphasis upon the objectives for the design and con
struction of each component.
Detector Matrix
The detectors utilized in the matrix were cubes of approximately
2 mm side length. The detectors were cut from large crystals
(2 mm x 2 mm x 7 mm) using a wire saw (0.003 diameter stainless steel
wire impregnated with 8 micron size diamonds). The crystal surfaces
were not treated (lapped or polished) after cutting (some of the
large crystal surfaces were not treated either). The cubes had
platinum electrodes applied to an area on two parallel opposed sides
(the electrodes did not extend to the edges of the crystals on the
large crystals or on the small cubes). The large detectors that were
cut to produce the small cubes were selected for the lowest noise,
best photon spectra and best physical characteristics (i.e., a
regular shape with electrode surfaces approaching crystal edges)
from approximately 130 available.
24

CdTe
Dtector
Matrix
Figure 3 Block Diagram Of X Y CdTe Camera

26
Substantial work was done in developing cutting techniques to
minimize degradation of spectrometer performance during cutting. Noise,
photon energy spectra and stability with time were examined for each
large crystal prior to cutting and for the small cubes after cutting.
Figure 4 represents a typical energy spectrum for a large crystal prior
to cutting and one of the small cubes after cutting (the noise voltage
and leakage current are also represented).
Wire leads were attached to the platinum electrodes with conducting
silver epoxy. Attempts were first made at mechanically coupling a wire
spring to the electrodes, but the hardware required could not be easily
manufactured. Attempts to use a silicon rubber based conducting
compound were also futile since the resultant bond between the wire and
the platinum was not adequate to allow handling of the crystal by its
wire leads. Ultrasonic bonding was also investigated but disregarded
since it requires high local temperatures and does not work well on
platinum surfaces. The wire used in ultrasonic bonding would not be
strong enough to allow for handling the crystal by its wire leads.
The epoxy utilized is a two component silver filled epoxy designed
specifically for chip bonding in microelectronic and optoelectronic
applications. It cures at temperatures as low as 50C (for 12 hours).
A detector, prior to wire lead bonding, is shown in figure 5 and after
wire lead bonding in figure 6.
A printed circuit board was designed that would facilitate the
coupling of rows to form X signals and the coupling of columns to form
Y signals (see figure 7). The board is a double sided printed circuit
board incorporating plated through holes. The wire leads of the detectors
were soldered to the printed circuit board. A layer of teflon was

COUNTS/CHANNEL COUNTS/CHANNEL
27
Detectors: CdTe
Isotope: Co-57
Bias: 100V
Preamplifier: Ortec 142A
Amplifier: Ortec 450(1.5^sec RCs)
Original Crystal Prior
To Cutting(ND035-6-3)
Leakage Current: 25nAs65V
One Of Small Crystals
After Cutting(ND035-6-3-A)
Leakage Current: 6nA965V
ENERGY I
122 KeV
Figure 4
Spectometry Performance Before And After Cutting

28
Figure 5 CdTe Detector (2x2x2 mm approximately)

29
Figure 6 CdTe Detector (With Leads Applied)

30
Figure 7 Printed Circuit Board

31
inserted between the detectors and the board for electrical insulation.
The printed circuit board has an edge connector that facilitates the
application of bias voltage and access to X and Y signals.
For convenience, the electrodes were oriented parallel to the
surfaces of the collimating holes. Ideally, the electrodes should be
oriented perpendicular to the collimating holes so that the negative
electrode is irradiated (as explained in Chapter 2). The operating
characteristics of each detector viere tested for both possible bias
voltage polarities. Each detector was ultimately operated in its
optimum bias voltage polarity.
Col 1imator
The research involved the design and fabrication of a multi hole
collimator. A multihole collimator was selected since it offers higher
sensitivity for a given resolution than the pinhole collimator, is useful
for large organ studies and is not subject to fall off in perimeter
response as is the pinhole collimator (Anger, 1975). A parallel hole
collimator was selected since the size of the image is essentially
independent of the distance from the subject to the collimator and for
a given resolution at a specified depth has better depth of focus (Anger,
1967). The most commonly used collimator is the parallel multihole
collimator. Although parallel multihole collimators were selected for
this research, the application of CdTe or other semiconductor radiation
detectors could be extended to focussed multi hole collimators and even
to pinhole collimators. For focussed collimators, the fabrication is
more difficult, but the in hole detector approach could still be utilized.
For pinhole collimators, the in hole detector approach could be applied

32
if the hole length were equal to the detector length so that scatter
between detectors would still be essentially eliminated.
The collimator holes are square since the cross section of the
available detectors is roughly square. CdTe cannot readily be made
into small diameter rods without inducing strains (ESPI, 1969). The
square collimator holes facilitated design calculations since septal
thickness is constant. The material selection for the collimator is
based on atomic number, density, machinability and expense. Tantalum
was used as the material due to its high Z (73), high specific gravity
(17.1), ability to be drawn into square tubes of appropriate dimensions
and relatively low expense. It is nontoxic. The collimator and
detector configuration were developed for use with the 140 keV photons
f 99mT
of Tc.
The square hole size was maintained at the minimum consistent with
the insertion of the detector and associated electrical leads and insula
tion. The detector electrodes were situated parallel to the hole axis.
If the electrodes had been perpendicular to the hole axis, the capacity
between the collimator and the detector electrodes would have been
minimized. If the electrodes are parallel to the hole axis, the
detector length and hence to some extent the sensitivity, could be
varied; although the capacitance may be increased and resolution de
creased.
Design criteria for parallel multi hole collimators as used in con
junction with an Anger camera are available (Anger, 1967; Causer, 1975).
The design criteria assume that the effective center of the detector is
located at some distance away from the back of the collimator. The

33
effective centers of the in hole detectors are located within the
collimator and hence existing design criteria are not exactly analagous.
Theoretical parallel multihole collimator designs attempt to optimize
the number of holes, their size, shape, septa thickness, and hole length.
Practical collimator design parameters have largely been assigned
empirically (My hill, 1967). Since the scope of this research was
primarily involved with the validity of the application rather than the
optimization of the collimator dimensions, selection of collimator dimen
sions was based on existing criteria for Anger type parallel hole col
limator design. Causer (1975) indicates that optimum collimator length
99m
is essentially independent of resolution and that for Tc photons,
optimum collimator length is approximately 1.8 2.5 cm. Collimator
length was selected as one inch since the effective collimator length
is reduced somewhat when the detector is located in hole. For a square
hole side dimension of about 0.115 inches, according to Anger (1967),
the septum thickness should be approximately 0.0081 inches. The actual
collimator dimensions are: length--1.0 inches (.01 inches); septa
thickness--0.008 inches ( 10%); square hole side dimension--0.115
inches ( 0.001 inches).
The shortest distance a photon can travel through a septum when
taking the unwanted path of minimum attenuation through the collimator
Q Q pi
is approximately 0.0355 inches. For Tc photons in tantalum this is
equivalent to an attenuation of 97%. The attenuation of scatter between
detectors for scattered photon energy of 140 keV is approximately 50%
and for scattered photon energy of 100 keV is approximately 80%.
The square tantalum tubes are assembled inside of an aluminum block
(see figure 8). Most of the tantalum tubes were larger than the square

34
Figure 8
Collimator

35
dimension tolerance and hence it was necessary to hand select tubes
and heat press them into the block. The component parts of the detecting
head (less detectors) are as shown in figure 9. A sleeve of insulating
tubing was placed around each detector when inserted into the collimator
so as to electrically insulate the detector from the tantalum tubes.
An aluminum foil was placed over the top of the collimator so as to
maintain the detectors in a dark environment (operation in light fields
increases the detector leakage currents). Additional layers of lead were
added to the exterior of the collimator to provide additional shielding.
Analog Circuitry
The analog portion of the electronics (i.e., that portion that
collects the charge from the rows and columns connecting the top electrodes
and bottom electrodes, respectively, and converts the charge to a voltage
representative of the energy deposited in a detection event) was
designed to meet the following criteria:
1) To produce a voltage pulse that is proportional in amplitude
to the charge pulse at the amplifier input and hence to the
energy deposited in a detection event.
2) To minimize the number and cost of components so as to
facilitate the hybridization that would be required for
clinically significant devices.
It is desirable to minimize the number of components since an n x n
matrix would require a minimum of 2n amplifiers. A clinical device
would require a large number of amplifiers and therefore in order to
minimize cost and to facilitate design hybridization, it is desirable to
use the fewest components commensurate with the application.

A B C D E F G
A-Base Plate
B-Spacer Block
C-Teflon Insulator
D-Printed Circuit Board
E-Teflon Insulator
F-Collimator
G-Cover Plate
Figure 9 Detector Head (Less Detectors)
36

37
In keeping with these ideals, a one stage amplifier that has
sufficient amplification to convert a relatively small charge pulse
collected from the crystals (on the order of less than 0.005 picocoulombs
for 100 percent charge collection from a 140 keV photon) to a voltage
pulse that has sufficient magnitude to be used directly in an analog to
digital conversion device was developed (see figure 10). It has been
reported that the actual charge collection efficiency in CdTe is sub
stantially less than 100% and is dependent upon electric field strength
and energy deposition location (Malm, 1973). The amplifier is essentially
a charge sensitive amplifier or charge to voltage converter. A primary
advantage of the charge sensitive design is that the output is independent
of detector capacity (which may vary with time and number of detectors
per row). The detector is alternating current (a.c.) coupled to the
preamplifier in order to remove the direct current (d.c.) component prior
to amplification. Direct current coupling would be preferred since it
is somewhat less noisy, but in this application, signals are acquired
from both of the detector electrodes which requires that at least one
side be a.c. coupled in order to remove the bias voltage component
applied to the detector. The amplifier consists of an operational
amplifier with a junction field effect transistor (JFET) at the input.
The resultant voltage output (V) is related to the charge input (Q) by
the expression:
V = Cf
where represents the capacity between the input and output of the
amplifier. In order to maximize the voltage output for a given input,
the capacity should be minimized, thus no added capacity was utilized

Figure 10 Analog Circuitry

39
between the input and output and the feedback capacity is the stray
capacitance between the input and output. A feedback resistor is
required to discharge the feedback capacity; otherwise, the feedback
capacity would build up a charge when input signals are present that
could eventually saturate the amplifier. A high value feedback resistor
is utilized in order to maintain a large time constant to maximize
the collection of charge deposited within the crystal. For the CdTe
detectors utilized in this research, an electric field of about 325 V/cm
gave optimum spectrometer performance. Since the electron and hole
2 2
mobilities of CdTe are on the order of 1000 cm /Vsec and 80 cm /Vsec
respectively, the charge collection time for a 325 V/cm field strength
is approximately 0.6 ysec for electrons and 8.0 ysec for holes. In
order to assure that the collection of positive charges (holes) and
negative charges (electrons) is maximized (in order to maximize the
amplifier output voltage per detection event), a 500 resistor was
utilized in the feedback network. The large feedback resistor thus assures
that the amplifier time constant is much much larger than the transit
time for hole collection and hence maximizes the charge collected per
detection event. The operational amplifier was selected for its high
gain and relatively low noise. The signals to be amplified occur at a
relatively low frequency (due to relatively long charge collection times
and due to a relatively low contemplated patient flux incident upon the
matrix), and hence the frequency-bandwidth characteristics of the
operational amplifier are not of primary importance. The JFET employed
at the input of the amplifier, determines the amplifier input impedance
and develops the signal/noise ratio that will be evident at the output
of the amplifier. The operational amplifier primarily provides the gain.

40
The placement of an external JFET at the input alleviates somewhat
the requirements for high input impedance for the amplifier. The
amplifier selected has relatively good specifications for all of these
characteristics but excels in gain.
Although many operational amplifiers have field effect transistors
at the input, an external junction field effect transistor was utilized
in this amplifier since they exhibit much lower noise figures than
insulated gate field effect transistors (metal-oxide-si 1 icon-field-
effect-transistor). The use of a discrete FET yields improved noise
characteristics in comparison to the fabrication in a monolithic design.
The particular JFET selected has a high common source forward
transconductance to common source input capacities ratio (Elad, 1969).
The ratio determines essentially the magnitude of change at the output
of the JFET per change at the JFET input. The JFET also has low noise.
A passive capacitor resistor network was added at the amplifier output
to differentiate the pulse to produce a pulse duration on the order of
50 ysec or less. The resultant pulse is bipolar but this is of little
consequence in this application since the output pulse is fed to the
analog to digital conversion device which is sensitive only to the
magnitude of the leading edge component of the pulse. The capacitor
resistor network also serves to remove the direct current voltage level
present at the output of the amplifier. A direct current voltage level
present at the output would complicate the design of the analog to digital
converter.
The power inputs to the amplifier are filtered by a resistor
capacitor network that serves to bypass ground loops and to smooth (by
integration) voltage fluctuations caused by wideband noise pickup.

41
This research utilized a 5 x 5 matrix of CdTe detectors and hence
required 10 amplifiers. The 10 amplifiers were fabricated onto one
printed circuit board and hence it was necessary to pay particular
attention to guarding the inputs of the amplifiers to minimize feedback
caused by voltage and current fluctuations in adjacent amplifiers.
A state of the art commercial preamplifier and amplifier were procured
for use in testing the CdTe detectors. The commercial amplifiers offer
low noise and large signal to noise ratio and the ability to control the
integration and differentiation of the signal. When using the commercial
amplifiers, the photopeak energy resolution is largely determined by the
noise generated within the detector. The commercial amplifiers were
used as a reference device for testing the detector properties and for
comparison of the inexpensive amplifiers developed for use with the
matrix.
As shown in figure 11 and figure 12, the low cost charge sensitive
amplifiers developed in this research offer signal to noise ratio and
photopeak resolution comparable to the state of the art commercial
amp!ifier.
Analog to Digital Circuitry
The analog to digital converter was designed to produce a digital
output signal whenever an input signal exceeds some reference level and
hence indicate that a detection event occurred. The reference level can
be adjusted to represent the lower limit or threshold of a photopeak and
hence pulse height discrimination can be performed. The digital outputs
of the analog to digital converter circuits are inputs to the digital

Pulse Height Pulse Height
42 .
Vv
MAUDERLI Waller -
429-72-4216
COMPARISON OF OUR SPECIALLY DESIGNED
PRE-AMPLIFIER WITH A COMMERCIALLY
AVAILABLE INSTRUMENT
Detector: CdTe
Isotope : Co- 57
ORTEC Pre-Amp.
Mod. 142-A
(followed byOrtec
Research Amp.
Mod 450)
Pre-Amp. Cost:
8 495.00
S/N ~ 6.5
Time
SPECIALLY DE
SIGNED Pre-Amp.
(followed by Or tec
Research Amp
Mod 450)
Cost of Material:
S 25.00
S/N ~6.3
cont. on next page wm 012078
Figure 1.1
Amplifier Signal-To-Noise Comparisons

COUNTS/CHANNEL COUNTS/CHANNEL
43
MAUDERLI Walter
429- 72-4216
Detector : CdTe
Isotope: Co-57
ORTEC Pre-Amp.
Mod. 142 A
(followed by Ortec
Research Amp
Mod. 450 and
Tracor Northern
Multichannel Analyzer
Mod.TN-1705
1024 Ch .)
SPECIALLY DE
SIGNED Pre-Amp.
(followed by the same
Instrumentsas above)
WM 012078
Figure 12 Amplifier Spectrum Comparisons

44
logic circuitry. The converter was also designed to facilitate hybridiza
tion. Since the output of amplifiers servicing rows is of opposite
polarity to the output of amplifiers servicing .columns, it was necessary
to develop converters for negative input signals and positive input
signals, that would produce a positive logic pulse when the input signal
exceeded some reference level.
The converters consist of voltage comparators as shown in figure 13.
The output of the comparators is fedback to the input in order to create
hysteresis and hence prevent input signal noise from causing the output
to switch states several times as the input signal passes through the
reference voltage level. Hysteresis serves to speed up the circuit
switching time and cause the comparator to switch only once as the input
signal passes through the reference voltage level and hence prevent
chatter. Chatter is deleterious to this application since it could
ultimately be evaluated as multiple detection events. The voltage
comparator was selected because it has a fast response time and reasonable
input offset and input bias currents. Capacitors were added from the
outputs of the comparators to ground in order to alleviate oscillations
caused by the output signal from one comparator being fedback by stray
coupling to the input of an adjacent comparator. The comparators were
assembled onto a printed circuit board but the inputs were inadequately
guarded to prevent the oscillations. The comparator printed circuit
board was shielded from the amplifier printed circuit board (and other
boards) in order to alleviate oscillation problems encountered due to
feedback caused by stray coupling.

150k
45
For Positive Analog Input Signals
(Voltage supplies are filtered as shown on figure 10.)
Analog
Input
For Negative Analog Input Signals
(Voltage supplies are filtered as shown on figure 10.)
Figure 13
Analog To Digital Converters

46
Digital Circuitry
The digital circuitry was developed to meet the following criteria:
1) To identify, based on the digital outputs of the row
and column circuitry the individual detector in which
a detection event occurs and to make that data
available for input to a microcomputer.
2) To eliminate from consideration, those events in which
more than one row signal or more than one column signal,
or both, occur within some coincident time frame.
The digital circuitry was designed such that the only events that
are collected as data are those in which one row signal and one column
signal occurred simultaneously (within some coincident time period) and
hence identify that a detection event occurred in a particular detector
(see figure 14). When a data input line goes high (after some quiescent
period when all inputs are low), the circuitry (after a brief wait for
coincidence), latches the data present. If the data consist of one X
signal and one Y signal, then that data are entered into a storage
device for subsequent transfer to a microcomputer port. If the data
are not acceptable or after completion of the data transfer, the process
starts again. The time frame in this particular device is not of
great importance since there are only 25 detectors supplying signals
and since the microcomputer requires a relatively long period of time
to input data.
The dual monostable multivibrators (one shots) A, B, C, D and E
serve to produce one 9.1 psec pulse each time that any of the XI X5
comparators or Y1 Y5comparators produce a transition from logic 0 to
logic 1 at the one shot input while the one shot output is low. The

PACKAGE
DESICIMAT10H
saimn
A.I.C.D.E.F.Y.HH CO 4098
Dual om shot
G.H.U.V
CD 4072
Dual 4 Input OR gat*
I,J,K,FF,GC
CD 4042
Quad docked D latch
L,M,AA,B8
CD 4030
Hex buffer
i,T
CD 4071
8 Input NOR gat*
O.P.Q.R
CD 4081
Quad 2 input AMD gate
S
CD 4011
Quad 2 Input HAND gat*
w,x
CD 4082
Dual 4 Input AMD gat*
CC.DO
3341
84 word-4 bit first in first
out serial eeaory(FIFO)
EE
m 7401
Quad 2 input HAND gat*
1M3084
Dlod*
Figure 14
Dig
ital Circuitry
(Page 1 of 2)

PACKAGE
DESTINATION
DESOtlPTIOH
A.B.C.D.E.F.Y.HH
CD 4098
Dual one shot
C.l.D.V
CD 4072
Dual 4 Input OK gate
I.J.K.FF.GC
CD 4042
Quad clocked D latch
L.M.AA.BB
CD 4050
Hex buffer
N.T
CD 4078
8 input NOR gate
O.P.Q.R
CD 4081
Quad 2 input AND gate
S
CD 4011
Quad 2 input NAND gate
w.x
CD 4082
Dual 4 input AND gate
CC.DD
3341
64 word-4 bit first in first
out serial wewory(FIFO)
EE
SN 7401
Quad 2 input NAND gate
-R-
IN3064
Diode
: 14
Digital
Circuitry (Page 2
OF 2)

I
49
9.1 ysec period of these one shots allow sufficient time for subsequent
data processing to be performed prior to the consideration of another
event. The one shots do not reset upon the application of each new
pulse and hence additional input pulses applied to a one shot during
the high period of the one shot will be ignored. The one shot outputs
are fed to OR gates G and H and to latches I, J, and K. When pin 1 of
OR gate G goes from low to high, one shot F enables (after a coincident
time delay of up to 5 ysec) the latches I, J, and K. The data present
at the latch inputs are held at the latch outputs for the 2.0 ysec of
the enable pulse (which allows time for subsequent data processing). It
is important that the only tim that one shot F will trigger is after
one shot A, B, C, D and E outputs have all been low and then one or
more low to high transitions start the analysis of another event.
Once the data have been latched in I, J, and K, they are buffered
through buffers L and M.
The output of the buffers is fed to AND gates 0, P, Q, R, OR gates
U and V, and NAND gate S, which in combination with NOR gates N and T
and AND gates W and X produce a high output at pin 1 of AND gate W if
data are present and the data consist of one X and one Y signal.
The outputs of buffers L and M are also fed to a diode matrix that
reduces the X data from 5 lines of decimal data to 3 lines of binary
data and likewise for the Y data. The diode matrix design could be
directly upgraded to a 16 x 16 matrix, by reducing 16 decimal lines to
4 binary lines on the X and Y axes and utilizing all 8 of the parallel
port input lines (rather than the 6 used in this application). When
pin 1 of AND gate W goes high, one shot Y is triggered, which enables
latches FF and GG to latch the data present on the X and Y binary data

50
lines. The output latches of FF and GG are buffered through buffer BB
to the inputs of first in first out serial memories (FIFO) CC and DD.
The FIFOs serve to assemble incoming data at one clock rate and transfer
the data out at a different clock rate. Each FIFO is a 64 word x 4 bit
storage device. Attempts to input data into the FIFO when all 64
registers are full are ignored. The FIFOs can easily be ganged to create
a composite FIFO of any dimension desired. When high, pin 1 of AND gate
W (after a delay through one shot AA for data set up) enables FIFOs CC
and DD to latch the data at their inputs. After data have been accepted
into the serial memory of the FIFOs, the FIFOs generate through NAND
gate EE a reset pulse that resets one shot Y to be ready for the next
data transfer.
Once data are entered in the FIFOs, data automatically "bubbles"
to the FIFO output. The microcomputer parallel input port (an eight
bit port) has two control signals that readily interface with the output
control lines of the FIFOs. The output ready lines of FIFOs CC and DD
activate NAND gate EE to create through buffer AA and one shot HH, a
ly sec pulse that pulls low the parallel data ready (PDR) line of the
microcomputer indicating that valid data are present at the port. The
microcomputer, under program control, inputs the data from the parallel
port and stores the data in memory for subsequent processing. Once
the data transfer into the microcomputer is complete, the parallel
input acknowledge line (PIAK) goes low to signal the FIFOs that the
data transfer is complete and that another data transfer may be initiated.
A master reset signal is activated by a switch, through one shot Y.
The FIFO memories are cleared and the FIFO control lines are initialized
when the master reset signal is activated.

51
The time frame for selection of a valid event is not very critical
in this application since the microcomputer uses in excess of 33 ysec
per data transfer. Complimentary metal oxide semiconductor (CMOS) devices
were utilized since they have low power consumption and good noise im
munity and since the data transfer time is not critical. Processing
time in the digital circuitry could be greatly reduced by employing
transistor-transistor logic (TTL) devices throughout and by incorporating
faster FIFOs.
Microcomputer
The microcomputer was selected to meet the following criteria:
1) To have sufficient hardware capabilities to readily input
data from a parallel port and store the data in memory with
minimal external interfacing.
2) To have compatible software available that can handle
moderately sophisticated tasks.
3) To be S-100 bus compatible.
4) To be based on a widely utilized microprocessor.
5) To have a cathode ray tube (CRT) terminal to facilitate
communication and programming.
6) To have the minimum cost consistent with 1 through 5 above.
The Sol 20 microcomputer, manufactured by Processor Technology
Incorporated was selected. It offers an Intel 8080 microprocessor
based microcomputer, uses the S-100 bus and provides both parallel and
serial standardized port interfaces with connectors, and an audio
cassette interface capable of controlling two recorders. It includes
a monitor program capable of operating programs written directly in

52
machine code. The microcomputer includes a serially encoded ASCII
keyboard and CRT that are useful not only in programming the micro
computer but also in utilizing the CRT terminal as a remote terminal
(via an acoustical coupler modem) to the local AMDAHL 470 system or
other compatible system. For more sophisticated programming require
ments, data can be transferred from the microcomputer memory to the
AMDAHL 470 and vice versa. A BASIC (Beginners All Purpose Symbolic
Instruction Code) program language that can be loaded in random access
memory (requiring approximately 5000 bytes of memory) is included. The
BASIC language has sufficient computing capability to operate the camera
built in this research. It facilitates transfer of program control
between BASIC to machine code programming.
A BASIC program was developed that allowed data from the parallel
port to be entered into memory. Each of the detectors was associated
with a particular memory location. Each time that data were entered from
the parallel port, that particular memory location was incremented and
the resultant sum examined to see whether a predetermined maximum count had
been achieved in that channel. The program was designed such that when
any one of the channels reached some predetermined count, data collection
would terminate. The program also included the capability of collecting
data from a flood or plane source and storing that in a separate memory
area for use in correcting for the individual sensitivities of the
detectors. Under program control the image memory or calibration memory
may be initialized (set to zero). The calibration matrix can be inverted
and multiplied by the image matrix, to develop a corrected image matrix.
The numerical elements of the calibration matrix and the corrected image
matrix can be displayed on the CRT.

53
The program also divides the image array into sixteen levels of
gray (coded 0 through 15) and formats the data in random access memory
for eventual use by a videographics interface. The sixteen gray levels
are derived by defining sixteen levels between 0 counts and the maximum
number of corrected counts per channel and assigning each matrix element
a value of 0 through 15 based on its corrected value.
Videographics Interface
The "Dazzler" videographics interface, manufactured by Cromemo
Incorporated, was utilized since it is the only interface readily avail
able that will read the random access memory of a S-100 compatible device
and display the data (properly formatted) in a 64 x 64 matrix with 16
shades of gray at each matrix location. The Dazzler does not require
servicing by the microprocessor during operation.
The Dazzler requires only 2048 bytes of computer memory. The
computer memory is scanned using direct memory access. The data in the
computer memory are formatted onto a TV screen to give a 64 by 64 matrix
with 16 shades of gray available at each matrix location. The Dazzler
output is a standard composite video signal that goes directly to a
monitor or television video amplifier. The microprocessor does encounter
some wait states caused by the Dazzler (while it reads the memory),
but this does not interfere with accurate program execution.
"In displaying the pictures, a compromise should be made between
contrast and gradation. Contrast should not be amplified far beyond
the point at which statistical variations become visible" (Anger, 1967,
p. 542). Gradation refers to the number of different gray tones used

54
and contrast refers to the ability to visibly distinguish between two
gray tones. Sixteen shades of gray are more than adequate in this
application, since brightness and contrast controls on the CRT may be
varied to derive the most meaningful image.
Summary
The camera described below was designed and constructed.
Twenty-five CdTe detectors were mounted on a printed circuit board
to form a 5 x 5 matrix. Each detector is located within an individual
hole of a collimator composed of individual square tubes of tantalum.
The collimator was designed for use with Tc-99m photons. The X and Y
outputs of the matrix are amplified in a one-stage amplifier and sub
sequently examined in a pulse height threshold device in an analog to
digital converter. Events representative of one X and one Y event
occurring within some coincident time frame, as judged by the digital
circuitry, are available for input to a microcomputer. A microcomputer
controls the input data, provides for image data collection, correction
and subsequent display via a videographics interface. The resultant
display is a 5 x 5 image consisting of variations of 16 shades of gray.

CHAPTER 5
RESULTS
The performance of the camera is best illustrated by images of
several pattern sources. Appropriate pattern source images, as recorded
by polaroid photographs of the video monitor, are presented in subsequent
figures. For comparison, images obtained with a conventional Anger
camera (Ohio Nuclear Series 100 Anger Camera) are also illustrated on
some of the figures.
A calibration matrix (a matrix of calibration factors obtained by
imaging a uniform plane source) is as shown in figure 15. The wide
variation evident in the matrix is discussed in the next chapter.
The resolution of the camera is illustrated for a point source in
figure 16. Figure 17 also demonstrates the resolution of the camera.
The two parallel lines are oriented along two rows of detectors with the
centers of the lines roughly aligned with the centerlines of the
detector rows. The resolution is essentially perfect in this configura
tion since adjacent rows represent the extremes of the gray scale.
Figure 18 is a representation of the same 1 ines rotated through approximately
45. The lines are centered around the central axis of the detector
matrix.
Figures 19 and 20 represent the limit of resolution for the matrix
constructed in this research. Figures 19 represents two equal strength
line sources that are just slightly further apart (3.5 mm) than the
physical dimension of one collimating tube (2.9 mm). Note that the
55

Source: Uniform plane (flood)
CALIBRATION DATA MATRIX
XI
X2
XI
H
2.00
1.34
6.29
J2
1.77
1.00
1.57
1
1.35
1.68
1.98
3£k
1.27
1.16
1.23
JE
1.29
2.11
2.71
Data not applicable¡detector removed
Figure 15 Calibration Data Matrix
m
X5
2.48
4.57
1.33
1.30
1.77
4096.00
1.48
1.36
1.53
2.65
DUE TO NOISE
cn
O'

i
57
Exposure Conditions:
Source: Tc-99m point source (1mm diameter)
Geometry: close to collimator
Scales: arbitrary and unequal
+
XY CdTe Cameba
-POINT CENTERED
OVER ONE DETECTOR
Anger Camera
-HIGH RESOLUTION
COLLIMATOR (1A0 KEV)
Figure 16 Point Source Images

Exposure Conditions
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: close to collimator
Scales: arbitrary and unequal
Activities: lines have equal specific activities
XY CdTe Camera
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Anger Camera
-HIGH RESOLUT
COLLIMATOR
KEV)
Figure 17 Parallel Line Source Images (6mm-Aligned)

Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: close to collimator
Scale: arbitrary
Activities: lines have equal specific activities
XY CdTe Camera
-lines rotated A5
W.R.T. ROWS
-LINE LENGTH DOES NOT EXCEED
SENSITIVE AREA
Figure 18 Parallel Line Source Image (6mm-Rotated)

Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter
3,5mm apart)
Geometry: close to collimator
Scales: arbitrary and unequal
Activities: lines have equal specific activities
ii I
XT CdTe Camera Anger Camera
-LINES ALIGNED -HIGH RESOLUTION
WITH COLUMNS COLLIMATOR (190 KEV)
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 19
Parallel Line Source Images (3.5mm-Aligned)

Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
2.5mm apart)
Geometry: close to collimator
Scales: arbitrary and unequal
Activities: specific activity of one line is twice
THAT OF THE OTHER LINE
I
XY CdTe Camera
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE AREA

-HIGH RESOLUTION
COLLIMATOR (1A0 KEV)
Figure 20 Parallel Line Source Images (2.5mm-Aligned)

62
center column does show some activity since the edges of both lines
overlap into the center column. Figure 20 illustrates that two line
sources of different activities can be imaged although the distance
between the lines (2.5 mm) is less than the physical dimension of one
collimating tube if the lines are oriented over two adjacent columns.
Degradation of image quality is observed by the introduction of 0.43
inch and 0.93 inch of Plexiglass scattering media between the lines
and the detector matrix is illustrated in figure 21 and figure 22. As
discussed subsequently, scattered radiation was not accurately discriminated
against. Geometry also contributes to the image degradation.
It was interesting to note during use of parallel line sources,
that air bubbles created within the device when the lines were loaded
could be diagnosed by the camera as shown in figure 23.
The sensitivity of the camera is qualitatively illustrated in
figure 24. The image was acquired in a counting time of approximately
3.8 seconds.
To illustrate the ability of the camera to image volume sources,
a thyroid phantom was imaged. The thyroid image represented in figure
25 is actually the composite of 36 images. Each image was normalized to
a common time frame by correcting for counting time and for decay of
the radioactivity. The 5 mm cold spot in the upper left hand corner is
somewhat spread out. This occurred because the cold spot was located at
the juncture of four composite images. Subsequent images of the cold
spot, with the cold spot centered over the detector matrix produced a
much sharper and better defined image.
The response of the camera was checked for paralysis by utilizing
a variety of uniform plane source activities (to a maximum of 0.146

63
mCi/cm^ of 99mTc). The counting rate was directly proportional to
source activity and no paralysis was evident.
The coincident time period was varied from near 0 to 5 yseconds
with no apparent change in image quality.
The background count rate was so low as to be inconsequential.
The X Y CdTe camera had substantial imaging capabilities as il
lustrated in the preceding figures. The results achieved as well as
difficulties encountered in achieving those results are discussed in
the next chapter.

64
Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: 0.43 inches of Plexiglas scattering media
IMPOSED BETWEEN COLLIMATOR AND SOURCE
Scale: arbitrary
Activities: lines have equal specific activities
XY CdTe Camera
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 21
Parallel Line Source Image (0.43 Inches
of Scattering Media)

Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
6mm apart)
Geometry: 0.93 inches of Plexiglas scattering media
IMPOSED BETWEEN COLLIMATOR AND SOURCE
Scale: arbitrary
Activities: lines have equal specific activities
XY CdTe Camera
-LINES ALIGNED
WITH ROWS
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 22
Parallel Line Source Image (0.93 Inches
of Scattering Media)

66
Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter,
5mm apart)
Geometry: close to collimatot
Scale: arbitrary
Activities: lines have equal specific activities (with
ONE AIR BUBBLE EVIDENT)
|
I
I
XY CdTe Camera
-LINES ALIGNED
WITH COLUMNS *BUBBLE (Air)
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 23 Parallel Line Source Image (Bubble Evident)

67
Exposure Conditions:
Source: Tc-99m parallel line source 1mm diameter,
5,5mm apart)
Geometry: close to collimator
Scale: arbitrary
Activities: left line-68/^Ci
RIGHT LINE-13S/UtCI
Imaging Time: 5.8 seconds
,i
XY CdTe Camera
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE VOLUME
Figure 24 Parallel Line Source Image (3.8 Seconds)

68
Exposure Conditions:
Source: Picker thyroid phantom (Tc-99m)
Geometry: close to collimator
Scales: arbitrary and unequal
XY CdTe Camera
Anger Camera
-composite of 36 -high resolution
images collimator (190 keV)
-pinhole collimator gave
equivalent results
Figure 25 Thyroid Phantom Images

CHAPTER 6
DISCUSSION OF RESULTS
The X Y CdTe camera concept offers the advantage of direct digital
localization of detected events, improvement of pulse height resolution
and attenuation of scatter within the detecting medium. As shown in the
previous chapter, the X Y CdTe camera produced images superior to other
existing devices. The images indicate that line source detection can
essentially be limited to the detectors directly beneath the line and
that imaging time is shorter than the other cameras tested. The per
formance of the X Y CdTe camera, however, was not without limitations.
The performance of the camera was limited by two important problems
1) The photopeak pulse heights of the individual CdTe detectors
varied by a factor of approximately two. The detector rows
were grouped according to pulse heights (i.e., each row had
detectors with similar pulse heights); however, as a result,
each detector column had a spectrum of photopeak pulse heights
ranging from very low to very high. Pulse height discrimina
tion for all columns was established at a level low enough
to accommodate detection of events in the detectors with very
small pulse heights. As a result, the detectors having
larger pulse heights produced a lot more counts than detectors
having small pulse heights.
2) Although at one time there were 25 detectors that exhibited
photopeaks, after a relatively short usage time, several of
69

70
the detectors became noisy when operated at a field strength
of 325 V/cm. In order to utilize the detector matrix the
bias voltage had to be reduced to 150 V/cm which precluded
spectrometry. When operated below 325 V/cm, the signal to
noise ratio is greatly reduced (particularly for those
detectors exhibiting small photopeak pulse heights initially)
and photopeak resolution capabilities are lost. The reason
for the noisy performance is essentially undetermined.
Throughout the research, a number of CdTe detectors failed
(exhibited no photopeak or became excessively noisy). The
crystal surfaces exposed in cutting were not lapped or
polished after cutting but then the original crystals were
not lapped and polished on all sides either (particularly the
ends of the crystals). The surface preparation done on the
original crystals was done by hand and was rather crude.
The effect of the two problems discussed above was that there was
a very large variation in detector sensitivity that could not be
balanced out by adjusting discrimination levels and elimination of
scatter events by pulse height discrimination was not possible (which made
elimination of scatter by the placement of detectors within the col
limating holes even more important).
The wide variation in detector sensitivities made conventional
analysis of plane sensitivity, modulation transfer function, line spread
function, etc., impractical due to the rather large statistical varia
tion in the less sensitive detectors. The variation in detector
sensitivities is illustrated in figure 15. The ability to measure line
spread function wasn't crucial; however, since as shown previously the

71
full width at half maximum is essentially the width of one matrix
element, and the resolving capabilities of the camera are rather easily
i 11ustrated.
The performance of the camera was compared to the rotational
parallel plate imaging device (PPRID) designed and constructed at the
University of Florida (Luthmann, 1978). Although the ultimate sensitivity
possible with either device has not been determined, it is possible to
establish that the PPRID has superior sensitivity to a conventional
Anger camera (see figure 26) since it can resolve two parallel lines of
75 yCi each in 9.4 sec. As shown in figure 24, the camera developed in
this research can resolve two lines of 138 yCi and 68 yCi in only 3.8
sec. Although the ability of both cameras to resolve two parallel lines
appears to be about the same, the resolution of the camera described
herein is somewhat superior in that the transition from the no-activity
regions to full-activity regions is absolute (i.e., goes immediately
and directly from the lightest gray scale to the darkest gray scale).
The PPRID camera produces a multitude of gray levels around and between
the two lines (not evident in figure 26 due to photographic technique),
such that the full width at half maximum encompasses more than the width
of one matrix element. Although both cameras seem to have about the
same ability to resolve parallel lines, thyroid phantom cold spots, etc.,
in all cases the image is somewhat superior for the camera described
herein since the contrast is superior. This would be expected since
the PPRID depends upon a computed reconstruction of the image from a
multitude of related images. The PPRID camera also has positioning
errors associated with it that do not have a direct analogy in this
camera. The radiologic image is directly measured by the X Y CdTe camera
described herein and hence should produce better images.

Front
Side
Lucife
DUAL LINE SOURCE
75 >jCi / Line
5 mm
0= Imm
iU
/Hole
/
1.5 mm =lmm ,
. Front \ S
U+ra-j
1
cr~+
u
L J
T UJ L
5mm
25 mm
5 5 mm
Lucite
1
Hole
12 mm
TRIPLE LINE SOURCE
27jjCi/Line
ROTATING LAMINAR CAMERA
ANGER CAMERA
Condition: Radioactive Lines at Face of Collimator
mm
Measuring time : 9.4 sec.
25 Profiles , 3/8 sec / Profile
Total counts: 9525
Measuring time: 18.9 sec.
Total counts: 10,000
0
Measuring time: 9.4 sec.
25 Profiles; 3/8 sec./Profile
Total counts : 6401
Measuring time: 15.6 sec.
Total count: 5000
Figure 26 PPRID and Anger Camera Comparisons

73
In summary, the X Y CdTe camera developed excellent resolution
and sensitivity. The direct digital localization was primarily
responsible for the success; however, in the absence of pulse height
resolution capabilities, the attenuation of scatter between elements in
the detector matrix became more important.

CHAPTER 7
CLINICAL FEASIBILITY
A clinically significant device should have a large enough sensitive
area to perform large organ studies or be large enough to be scanned or
composited to assimilate large organ images. Anger cameras have sensitive
2
areas as large as approximately 1000 cm Some of the considerations
for a clinical device are discussed below.
1) The matrix of a clinical device could be composed of sub
groups. This is dependent upon the number (or volume) of
CdTe detectors that can be operated in a row without
serious degradation of signal to noise ratio or pulse
height resolution. During testing of the small crystals
after cutting operations, it was demonstrated that at least
five of the small crystals could be operated in a row.or column
while still maintaining a distinguishable photopeak. The
manufacturer of CdTe has demonstrated that four to eight
detectors for a total aggregate volume of approximately
2 x 2 x 50 mm can be operated and still maintain a dis
tinguishable photopeak. The photopeak resolution of the
aggregate is substantially worse than any of the individual
detectors due to statistical signal to noise degradation
and due to the fact that no two detectors have essentially
the same photopeak pulse height which requires that the
detectors be roughly grouped according to pulse height
74

75
magnitude and according to noise. It should be noted
that the noise of the original crystals available varied
from 25 to 200 nA.
2) The size of a subgroup would essentially be determined by
the properties of the detectors available (i.e., limited
by noise and photopeak resolution considerations for
crystals in rows). A microcomputer system with two 8-bit
parallel input ports could easily handle a 256 x 256 matrix.
A 256 x 256 matrix would be adequate for clinical devices (even
at 1 mm x 1 mm element dimensions).
3) The size of the crystals and associated, insulation and
shielding between detectors could be reduced to produce
finer resolution. The ultimate size limitation depends upon
the ability of a manufacturer to mass produce small
crystals with consistent detection properties. A reduction
in the size of a matrix element requires an attendant increase
in associated electronics.
4) The crystals could ultimately be mounted directly on a
silicon substrate which could facilitate alignment of the
crystals, handling of large numbers of crystals and lead wire
attachment. Orientation could be established such that the
negative electrode is irradiated so as to take full advantage
of electron mobility (in comparison to hole mobility).
5) The concept of placing attenuating media between detectors
could be altered such that narrow strips of shielding could
be placed between the detectors rather than inserting the
detectors into the actual collimator tubes. This would enable

76
the ability to use several collimator designs (as is done
on the Anger camera) without sacrificing the absorption of
scatter radiation between detectors.
6) The analog portion of the circuitry could be redesigned to
make the voltage output independent of the position of charge
deposition (and hence increase the proportion of detections
indicative of full energy photon detection). By utilizing
a circuit time constant that is much much shorter than the
collection time for electrons, the slope of the output
pulse (dV/dt) would be proportional to the number of ion
pairs produced and the magnitude of the output becomes:
V = R(dq/dt).
The output voltage pulse is proportional to the current flow
and hence its height is a measure of the number of ion pairs
created (Price, 1964). The closed loop gain of an amplifier
with such a short circuit time constant would be very low and
hence a second amplifier stage might be required. Design
considerations for low noise second stage amplifiers are
available (Arbel, 1968). Alternately, a precision microvolt
comparator could be used. The use of a very short circuit
time constant would facilitate higher sensitivity (higher
count rate) since pulse duration would be shorter. The
performance of the detectors and the performance of the as
sociated analog circuitry could be improved by cryogenic
operation (Elad, 1969).
7) Means for adjusting the timing of each analog to digital
electronics circuit would be useful in order to synchronize
the response to a given input pulse.

77
8) High speed TTL FIFOs are currently available that input
clock rate capabilities in excess of 10 MHZ. By ganging
large numbers of these relatively inexpensive memories, it
would be possible to store data as fast as the detector
matrix produces it and hence maximize camera sensitivity.
The data could later be transferred into computer memory at
the much slower rate required for that data transfer.
9) The image might be improved by inserting interpolated data
between actual detector measured data, based on the average
of measured data from adjacent elements. This could tend to
smooth out the transition of data between measured data
elements. An alternate gray level calculation scheme could
also be used. Rather than simply identifying sixteen levels
between zero and the maximum count obtained, it is possible
to define sixteen gray levels between the minimum count and
maximum count or base the image around an average count.
These procedures will not alter the basic data upon which the
image is based, but may produce an image that is aesthetically
more pleasing or easier to visualize. It should also be noted
that color images can also be produced.
The basic clinical feasibility of a camera based on the
concept of constructing a large matrix of small detectors depends
upon the availability of large numbers of small detectors that
have essentially the same radiation detection properties. The
variability that exists, even in specially selected or grouped
CdTe detectors, is too large to fabricate a CdTe matrix camera

78
at this time. The recent availability of intrinsic germanium
detectors (1 mm x 1 mm x 7 mm) might make a large camera
feasible. However, the uniformity of radiation detection
properties and the pulse height resolution of a large series
of the detectors is unknown.
The direct digital localization of detected events and
the attenuation of scatter radiation in the detector plane
offer great promise for clinical devices. The availability
of large numbers of small detectors with uniform detection
characteristics would make the camera clinically feasible.

BIBLIOGRAPHY
Allemand R, Bouteiller P, Laval M: Present limitations of CdTe
detectors in nuclear medicine. Rev Phys Appl 12:365-367, 1977
Anger HO: Radioisotope cameras. Instrumentation in Nuclear Medicine.
New York, Academic Press, 1967, pp 485-552
Anger HO: Instruments: specific devices. Nuclear Medicine. New York,
HP Publishing Co, 1975, pp 29-40
Arbel FA: The second stage noise contribution of a nuclear pulse
amplifier. IEEE Trans Nucl Sci 15:2-5, 1968
Armantrout GA, Swierkowski SP,. Sherohman JW, Yee JH: What can be
expected from high z semiconductor detectors? IEEE Trans Nucl Sci
24:121-125, 1977
Aven M, Prener JS, Editors: Physics and Chemistry of 11-VI Compounds.
Amsterdam, North Holland Publishing Co, 1967
Beck RN: Instruments: basic principles. Nuclear Medicine. New York,
HP Publishing Co, 1975, pp 17-28
Causer DA: The design of parallel hole gamma camera collimators. Int J
Appl Radi at Iso 26:355-362, 1975
Detko JF: A prototype ultra pure germanium orthogonal strip gamma
camera. Med Radioisotopes Scintigraphy 1:241-254, 1972
El ad E: FET preamplifiers for semiconductor radiation detectors.
ISPRA Nuclear Electronics Symposium Proceedings, 1969 pp 21-34
Entine G, Garcia DA, Tow DE: Review of CdTe medical applications. Rev
Phys Appl 12:355-359, 1977
ESPI Catalog 690. Los Angeles, Electronic Space Products Incorporated
1969, p 181
Hoffer PB, Beck RN, Gottschalk A, Editors: Semiconductor Detectors in the
Future of Nuclear Medicine. New York, The Society of Nuclear
Medicine, 1971, p 287
Hubbell JH: Photon Cross Sections, Attenuation Coefficients, and Energy
Absorption Coefficient From 10 keV to 100 GeV. Washington,
National Bureau of Standards, 1969, NSRDS-NBS29
79

80
Luthmann RW: Rotational parallel plate radionuclide imaging device.
Doctoral Dissertation, University of Florida, 1978
Malm HL, Raudorf TW, Martini M, Zanio KR: Gamma ray efficiency
comparisons for Si(Li), Ge, CdTe and Hgl0 detectors. IEEE Trans
Nucl Sci 20:500-509, 1973
Meyer E, Desaulniers G, Sternberg J: The CdTe subminiature semiconductor
detector: a versatile instrument for detection of gamma radiation.
Int J Appl Radiat Iso 26:697-702, 1975
Moretti JL, Mensch B, Guey A, Desgrey A: Comparative Assays of Scintil-
lation Cameras (Anger Type). Cleveland, Nuclear Medicine Institute,
1974
My hill J, Hie GJ: Multi hole collimators for scanning. Instrumentation
in Nuclear Medicine, New York, Academic Press, 1967, pp 429-460
Palms JM: Newer developments in detector design and materials.
Semiconductor Detectors in the Future of Nuclear Medicine.
New York, The Society of Nuclear Medicine Inc., 1971, pp 58-77
Price WJ: Nuclear Radiation Detection. New York, McGraw-Hill Book
Company, 1964
Serreze HB, Entine G, Bell R0, Wald FV: Advances in CdTe gamma ray
detectors. IEEE Trans Nucl Sci 21:404-407, 1974
Siffert P, Cornet A, Stuck R, Triboulet R, Marfaing Y: Cadmium telluride
nuclear radiation detectors. IEEE Trans Nucl Sci 22:211-225, 1975
Siffert P, Berger J, Scharager C, Cornet A, Stuck R, Bell R0,
Serreze HB, Wald FV: Polarization in cadmium telluride nuclear
radiation detectors. IEEE Trans Nucl Sci 23:159-170, 1976
Sveinsdottir E, Larsen B, Rommer P, Lassen NA: A multidetector scintil
lation camera with 254 channels. J Nucl Med 18:168-174, 1977
Wald FV: Applications of CdTe, a review. Rev Phys Appl 12:277-290, 1977
Zanio K: Use of various device geometries to improve the performance
of CdTe detectors. Rev Phys Appl 12, 365-367, 1977

BIOGRAPHICAL SKETCH
Jerry David Allison was born on June 18, 1948, in Brevard, North
Carolina. He attended local primary and elementary schools and entered
North Carolina State University in 1966. He was graduated in 1970 with
a Bachelor of Science in nuclear engineering. Mr. Allison was employed
at Newport News Shipbuilding in Virginia as a Radiological Control Engineer
from 1970 until 1975. In 1974, he earned the Master of Engineering
degree in mechanical engineering from Old Dominion University in Norfolk.
In 1975, he entered the University of Florida for graduate work in
medical radiation physics.
Mr. Allison is a licensed Professional Engineer in the Commonwealth
of Virginia and is certified in health physics by the American Board
of Health Physics.
He is married to the former Mary Jacqueline Andrews, also of
Brevard. The Allisons have two sons.
81

I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
\
7
Ci\:>
ICCs- J rCl'
Walter Mauderli, Chairman
Professor of Nuclear Engineering
Sciences
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
W. Emmett Bolch
Professor of Nuclear Engineering
Sciences
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
Valerie A Brookeman
Associate Professor of Radiology
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
!ojx;
LV
/ /
(A.-l-
Lawrence T. Fitzgerald
Assistant Professor'of Nuclear
Engineering Sciences

I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
/
cls-'x -c\r-c
Genevieve S. Roessler
Assistant Professor of Nuclear
Engineering Sciences
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
Henri A. Van Rinsvelt
Professor of Physics
This dissertation was submitted to the Graduate Faculty of the College
of Engineering and to the Graduate Council, and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
August 1978
/¡tJhd &
Dean, College of Engineering
Dean, Graduate School

UF Libraries: Digital Dissertation Project
6/17/08 9:2.1 AM
Dear Dr. Jerry D. Allison,
The George A. Smathers Libraries at the University of Florida has initiated a project to retrospectively digitize and
make available on the Internet any dissertation written by a University of Florida doctoral candidate and accepted by
the University of Florida. It is our goal to make the documents fully text searchable and easily harvested by Internet
search engines, allowing the full breadth and scope of scholarship produced at the University of Florida to be made
available across the world quickly and easily. The Library is bearing the full cost of the project.
We would like to add your dissertation, The application of cadmium telluride as in hole semiconductor radiation
detectors in an X Y matrix radioisotope camera /, published in 1978, to the project. In order to do so we need a
signed, legal original Internet Distribution Consent Agreement for our files. If you want your dissertation included in
the project, please print the Consent Agreement on the page below, sign it and mail it back to the Libraries at the
address listed.
We will keep you informed of the progress of your dissertation as it works its way through the project.
If you have any questions, please reply to UFdissertations@uflib.ufl.edu.
Thank you,
Cathy Martyniak, Project Coordinator
Christy Shorey, Project Technician
fin UNIVERSITY of
Ur FLORIDA
The Foundation for The Gator Nation
Internet Distribution Consent Agreement
In reference to the following dissertation:
AUTHOR: Allison, Jerry
TITLE: The application of cadmium telluride as in hole semiconductor radiation
detectors in an X Y matrix radioisotope camera / (record number: 74791)
PUBLICATION DATE: 1978
I, 1^as copyright holder for the aforementioned dissertation, hereby grant
specifte'ap'd limited aryhiVe and distribution rights to the Board of Trustees of the University of Florida and its agents. I authorize
the Ifnjtfersity of Florida to digitize and distribute the dissertation described above for nonprofit, educational purposes via the
Internet or successive technologies.
This is a non-exclusive grant of permissions for specific off-line and on-line uses for an indefinite term. Off-line uses shall be
limited to those specifically allowed by "Fair Use" as prescribed by the terms of United States copyright legislation (cf, Title 17,
U.S. Code) as well as to the maintenance and preservation of a digital archive copy. Digitization allows the University of Florida
to generate image- and text-based versions as appropriate and to provide and enhance access using search software.
This gp
igitized versions for commercial use or profit.
Imain:
Page 1 of 2

6/17/08 9:2.1 AM
UF Librangs: Digital Dissertation Project
S)C\r\M/ /-)
Pk i
D
Printer! nr Tvnerl Nafne nf tloWriffht Hnlder/Licensee
A
Personal information blurred
Hal&3
Please print, sign and return to:
Cathleen Martyniak
UF Dissertation Project
Preservation Department
University of Florida Libraries
P.O. Box 117007
Gainesville, FL 32611-7007
mhtmlmain:
Page 2 of 2



TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS Hi
ABSTRACT v
CHAPTER
1. INTRODUCTION 1
2. HISTORICAL DEVELOPMENT 3
3. CAMERA PROPERTIES 7
Advantages of a CdTe Matrix 16
Advantages of In Hole Detectors 19
Disadvantages 23
4. CAMERA DESIGN AND CONSTRUCTION 24
Detector Matrix 24
Collimator 31
Analog Circuitry 35
Analog to Digital Circuitry- 41
Digital Circuitry 46
Microcomputer 51
Videographics Interface 53
Summary 54
5. RESULTS 55
6. DISCUSSION OF RESULTS 69
7. CLINICAL FEASIBILITY 74
BIBLIOGRAPHY 79
BIOGRAPHICAL SKETCH
81


8
and incoherent) are useful (excluding the positron camera principle).
The origin of scattered photons detected by a camera cannot be limited
to a solid angle as is possible with unscattered photons. In order to
eliminate some of the signals generated by photons that have been in
coherently scattered (and hence have less energy), a pulse height
analyzer is incorporated in conventional cameras. The pulse height
analyzer discriminates against those signals whose amplitude is not
large enough to indicate that a full energy photon (i.e., a monoenergetic
photon whose energy has not been reduced by scatter) was absorbed by
the crystal. In practice, it is not possible to precisely discriminate
against scattered photons since even full energy photon pulse heights
exhibit statistical variation about some mean value. The signal caused
by a scattered photon or the coincident summation signal caused by
several scattered photons may have an amplitude within the statistical
variation of full energy photon signals. It is possible to discriminate
against some of the scattered photons by setting a threshold pulse
height that is just below the photopeak of the full energy photons.
Coherently scattered photons are essentially full energy photons that
are scattered into some solid angle without significant energy transfer
occurring. Coherently scattered photons do not contribute useful
informaton to the image formation process but cannot be discriminated
against. A collimator is incorporated in the Anger camera to create a
spatial relationship between the image created by full energy absorp
tion events in the crystal and the distribution of radioactivity in the
object.
"Semiconductor detectors offer many advantages for clinical and
experimental uses, chiefly because of their small size and excellent


40
The placement of an external JFET at the input alleviates somewhat
the requirements for high input impedance for the amplifier. The
amplifier selected has relatively good specifications for all of these
characteristics but excels in gain.
Although many operational amplifiers have field effect transistors
at the input, an external junction field effect transistor was utilized
in this amplifier since they exhibit much lower noise figures than
insulated gate field effect transistors (metal-oxide-si 1 icon-field-
effect-transistor). The use of a discrete FET yields improved noise
characteristics in comparison to the fabrication in a monolithic design.
The particular JFET selected has a high common source forward
transconductance to common source input capacities ratio (Elad, 1969).
The ratio determines essentially the magnitude of change at the output
of the JFET per change at the JFET input. The JFET also has low noise.
A passive capacitor resistor network was added at the amplifier output
to differentiate the pulse to produce a pulse duration on the order of
50 ysec or less. The resultant pulse is bipolar but this is of little
consequence in this application since the output pulse is fed to the
analog to digital conversion device which is sensitive only to the
magnitude of the leading edge component of the pulse. The capacitor
resistor network also serves to remove the direct current voltage level
present at the output of the amplifier. A direct current voltage level
present at the output would complicate the design of the analog to digital
converter.
The power inputs to the amplifier are filtered by a resistor
capacitor network that serves to bypass ground loops and to smooth (by
integration) voltage fluctuations caused by wideband noise pickup.


13
the mean free path of the charge carriers should be less than the col
lection distance or detector width. For the best photopeak energy
resolution in a 2 mm detector thickness, the bias voltage applied should
be on the order of 500 volts. The detector generated noise due to
increased leakage current caused by electric fields as high as 2500 V/cm
make operation impractical at that level. It is recommended that the
bias voltage be maintained at the maximum practical voltage (Malm,
1973). For a practical bias voltage of only 500 V/cm, the charge col
lection time for a 2 mm detector would range from approximately 0.4 ysec
to 5 ysec. The decay time of a Nal(Tl) scintillator is approximately
0.3 ysec.
The detector thickness is essentially limited by the field strength.
Reasonable spectrometer performance for the CdTe available occurs
around 325 to 500 V/cm. For a minimum trapping time of 6 ysec and
2
mobilities of 80 and 1000 cm /V-sec for holes and electrons respectively,
the detector thickness is limited to about 2 mm. For higher field
strengths, noise degrades the spectrometer performance. Further
development of CdTe could enable higher spectrometer field strengths
and hence thicker detectors.
Cadmium telluride has many other properties which are not of great
interest in semiconductor radiation detector applications; however,
for completeness some of those properties are listed as follows:
piezoelectric, photovoltaic, magnetoresistive, pyroelectric, electro
optic (Aven, 1967). A listing of the main applications of CdTe is as
follows: infrared windows, electro-optic modulator, Gunn effect high
frequency current oscillator, piezoelectric devices, luminescent devices,
solar cells and nuclear detectors (Wald, 1977).


6
separated by lead strips in order to reduce the detection
of scattered radiation. The quantascope utilized an
image intensifier rather than PMTs. The camera produced
poorer contrast than is available with an Anger camera and
required matched crystal selection (Beck, 1975).
4) Another CdTe radioisotope camera was recently constructed;
however, the application involves computer-assisted recon
struction techniques for the development of a two dimensional
representation of the three dimensional radioisotope distri
bution (Luthmann, 1978). Data is obtained from the n detector
rows situated within a parallel plate collimator which is
rotated about its central axis.
Most of the radioisotope cameras that have been developed, offer
improvement in some parameter as compared to the Anger camera. A com
plete evaluation of the devices reveals that a multipurpose device that
can offer substantial improvement in sensitivity, spatial and temporal
resolution and hence an improvement in diagnostic information obtained
or a reduction in patient exposure, has not been developed.


Front
Side
Lucife
DUAL LINE SOURCE
75 >jCi / Line
5 mm
0= Imm
iU
/Hole
/
1.5 mm =lmm ,
. Front \ S
U+ra-j
1
cr~+
u
L J
T UJ L
5mm
25 mm
5 5 mm
Lucite
1
Hole
12 mm
TRIPLE LINE SOURCE
27jjCi/Line
ROTATING LAMINAR CAMERA
ANGER CAMERA
Condition: Radioactive Lines at Face of Collimator
mm
Measuring time : 9.4 sec.
25 Profiles , 3/8 sec / Profile
Total counts: 9525
Measuring time: 18.9 sec.
Total counts: 10,000
0
Measuring time: 9.4 sec.
25 Profiles; 3/8 sec./Profile
Total counts : 6401
Measuring time: 15.6 sec.
Total count: 5000
Figure 26 PPRID and Anger Camera Comparisons


31
inserted between the detectors and the board for electrical insulation.
The printed circuit board has an edge connector that facilitates the
application of bias voltage and access to X and Y signals.
For convenience, the electrodes were oriented parallel to the
surfaces of the collimating holes. Ideally, the electrodes should be
oriented perpendicular to the collimating holes so that the negative
electrode is irradiated (as explained in Chapter 2). The operating
characteristics of each detector viere tested for both possible bias
voltage polarities. Each detector was ultimately operated in its
optimum bias voltage polarity.
Col 1imator
The research involved the design and fabrication of a multi hole
collimator. A multihole collimator was selected since it offers higher
sensitivity for a given resolution than the pinhole collimator, is useful
for large organ studies and is not subject to fall off in perimeter
response as is the pinhole collimator (Anger, 1975). A parallel hole
collimator was selected since the size of the image is essentially
independent of the distance from the subject to the collimator and for
a given resolution at a specified depth has better depth of focus (Anger,
1967). The most commonly used collimator is the parallel multihole
collimator. Although parallel multihole collimators were selected for
this research, the application of CdTe or other semiconductor radiation
detectors could be extended to focussed multi hole collimators and even
to pinhole collimators. For focussed collimators, the fabrication is
more difficult, but the in hole detector approach could still be utilized.
For pinhole collimators, the in hole detector approach could be applied


32
if the hole length were equal to the detector length so that scatter
between detectors would still be essentially eliminated.
The collimator holes are square since the cross section of the
available detectors is roughly square. CdTe cannot readily be made
into small diameter rods without inducing strains (ESPI, 1969). The
square collimator holes facilitated design calculations since septal
thickness is constant. The material selection for the collimator is
based on atomic number, density, machinability and expense. Tantalum
was used as the material due to its high Z (73), high specific gravity
(17.1), ability to be drawn into square tubes of appropriate dimensions
and relatively low expense. It is nontoxic. The collimator and
detector configuration were developed for use with the 140 keV photons
f 99mT
of Tc.
The square hole size was maintained at the minimum consistent with
the insertion of the detector and associated electrical leads and insula
tion. The detector electrodes were situated parallel to the hole axis.
If the electrodes had been perpendicular to the hole axis, the capacity
between the collimator and the detector electrodes would have been
minimized. If the electrodes are parallel to the hole axis, the
detector length and hence to some extent the sensitivity, could be
varied; although the capacitance may be increased and resolution de
creased.
Design criteria for parallel multi hole collimators as used in con
junction with an Anger camera are available (Anger, 1967; Causer, 1975).
The design criteria assume that the effective center of the detector is
located at some distance away from the back of the collimator. The


To My Parents
a£


54
and contrast refers to the ability to visibly distinguish between two
gray tones. Sixteen shades of gray are more than adequate in this
application, since brightness and contrast controls on the CRT may be
varied to derive the most meaningful image.
Summary
The camera described below was designed and constructed.
Twenty-five CdTe detectors were mounted on a printed circuit board
to form a 5 x 5 matrix. Each detector is located within an individual
hole of a collimator composed of individual square tubes of tantalum.
The collimator was designed for use with Tc-99m photons. The X and Y
outputs of the matrix are amplified in a one-stage amplifier and sub
sequently examined in a pulse height threshold device in an analog to
digital converter. Events representative of one X and one Y event
occurring within some coincident time frame, as judged by the digital
circuitry, are available for input to a microcomputer. A microcomputer
controls the input data, provides for image data collection, correction
and subsequent display via a videographics interface. The resultant
display is a 5 x 5 image consisting of variations of 16 shades of gray.


Exposure Conditions:
Source: Tc-99m parallel line source (1mm diameter
3,5mm apart)
Geometry: close to collimator
Scales: arbitrary and unequal
Activities: lines have equal specific activities
ii I
XT CdTe Camera Anger Camera
-LINES ALIGNED -HIGH RESOLUTION
WITH COLUMNS COLLIMATOR (190 KEV)
-LINE LENGTH EXCEEDS
SENSITIVE AREA
Figure 19
Parallel Line Source Images (3.5mm-Aligned)


Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
THE APPLICATION OF CADMIUM TELLURIDE AS IN HOLE SEMICONDUCTOR
RADIATION DETECTORS IN AN X Y MATRIX RADIOISOTOPE CAMERA
By
Jerry David Allison
August 1978
Chairman: Walter Mauderli, D. Sc.
Major Department: Nuclear Engineering Sciences
A radioisotope camera that has superior performance in comparison
to conventional devices, has been designed, constructed, and tested.
Radioisotope cameras are instruments used for imaging the distribution
of radioisotopes in vivo. A camera creates a two dimensional image of
the three dimensional radioisotope distribution within an object. Con
ventional radioisotope cameras are mainly of the Anger type. The radia
tion detector in an Anger camera is a sodium iodide scintillation crystal
that is located in back of a collimator. The collimator creates a
spatial relationship between the image and the radioisotope distribution.
The new camera is a small experimental camera that has a matrix of
cadmium telluride semiconductors as the detecting medium. The detectors
are located within the collimating holes (rather than behind them).
Coincident outputs from each row and column are used to digitally
identify the individual detector in which the energy deposition occurred.
Direct digital localization of detected events and the attenuation of
v


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
/
cls-'x -c\r-c
Genevieve S. Roessler
Assistant Professor of Nuclear
Engineering Sciences
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.
Henri A. Van Rinsvelt
Professor of Physics
This dissertation was submitted to the Graduate Faculty of the College
of Engineering and to the Graduate Council, and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
August 1978
/¡tJhd &
Dean, College of Engineering
Dean, Graduate School


I
49
9.1 ysec period of these one shots allow sufficient time for subsequent
data processing to be performed prior to the consideration of another
event. The one shots do not reset upon the application of each new
pulse and hence additional input pulses applied to a one shot during
the high period of the one shot will be ignored. The one shot outputs
are fed to OR gates G and H and to latches I, J, and K. When pin 1 of
OR gate G goes from low to high, one shot F enables (after a coincident
time delay of up to 5 ysec) the latches I, J, and K. The data present
at the latch inputs are held at the latch outputs for the 2.0 ysec of
the enable pulse (which allows time for subsequent data processing). It
is important that the only tim that one shot F will trigger is after
one shot A, B, C, D and E outputs have all been low and then one or
more low to high transitions start the analysis of another event.
Once the data have been latched in I, J, and K, they are buffered
through buffers L and M.
The output of the buffers is fed to AND gates 0, P, Q, R, OR gates
U and V, and NAND gate S, which in combination with NOR gates N and T
and AND gates W and X produce a high output at pin 1 of AND gate W if
data are present and the data consist of one X and one Y signal.
The outputs of buffers L and M are also fed to a diode matrix that
reduces the X data from 5 lines of decimal data to 3 lines of binary
data and likewise for the Y data. The diode matrix design could be
directly upgraded to a 16 x 16 matrix, by reducing 16 decimal lines to
4 binary lines on the X and Y axes and utilizing all 8 of the parallel
port input lines (rather than the 6 used in this application). When
pin 1 of AND gate W goes high, one shot Y is triggered, which enables
latches FF and GG to latch the data present on the X and Y binary data


CHAPTER 5
RESULTS
The performance of the camera is best illustrated by images of
several pattern sources. Appropriate pattern source images, as recorded
by polaroid photographs of the video monitor, are presented in subsequent
figures. For comparison, images obtained with a conventional Anger
camera (Ohio Nuclear Series 100 Anger Camera) are also illustrated on
some of the figures.
A calibration matrix (a matrix of calibration factors obtained by
imaging a uniform plane source) is as shown in figure 15. The wide
variation evident in the matrix is discussed in the next chapter.
The resolution of the camera is illustrated for a point source in
figure 16. Figure 17 also demonstrates the resolution of the camera.
The two parallel lines are oriented along two rows of detectors with the
centers of the lines roughly aligned with the centerlines of the
detector rows. The resolution is essentially perfect in this configura
tion since adjacent rows represent the extremes of the gray scale.
Figure 18 is a representation of the same 1 ines rotated through approximately
45. The lines are centered around the central axis of the detector
matrix.
Figures 19 and 20 represent the limit of resolution for the matrix
constructed in this research. Figures 19 represents two equal strength
line sources that are just slightly further apart (3.5 mm) than the
physical dimension of one collimating tube (2.9 mm). Note that the
55


Source: Uniform plane (flood)
CALIBRATION DATA MATRIX
XI
X2
XI
H
2.00
1.34
6.29
J2
1.77
1.00
1.57
1
1.35
1.68
1.98
3£k
1.27
1.16
1.23
JE
1.29
2.11
2.71
Data not applicable¡detector removed
Figure 15 Calibration Data Matrix
m
X5
2.48
4.57
1.33
1.30
1.77
4096.00
1.48
1.36
1.53
2.65
DUE TO NOISE
cn
O'


18
since the output is an address developed by the detector in
which the energy is deposited. The detectors are separated
from each other by the collimator septa and hence the probability
that a photon can deposit energy in more than one detector
is substantially reduced.
3) Improved photopeak resolution is possible with CdTe. In
scintillation detectors, the deposition of energy in the crystal,
by radiation, results in ionization and excitation. The energy
is converted to light by the luminescence process. The light
travels through the crystal and by means of a light pipe and
reflectors is transmitted to the photocathode of the PMT. The
light is absorbed by the photocathode resulting in photo
electron emission. An electron multiplication process
produces a large pulse at the output of the PMT. Each step
in the process has its own efficiency and statistical varia
tion. The statistical variation of the PMT output pulse is
the propagation of the variations of the individual steps
involved.
The semiconductor radiation detection principle of CdTe
is somewhat simpler in that the ion pairs (created by energy
deposition) are collected at the electrodes due to the electric
field developed across the detector by the bias voltage
(assuming trapping and recombination are prevented). The con
version of ion pairs to light and light to electrons and the
associated statistical variations are eliminated. The charge
collected at the electrodes is amplified into a large pulse.


Pulse Height Pulse Height
42 .
Vv
MAUDERLI Waller -
429-72-4216
COMPARISON OF OUR SPECIALLY DESIGNED
PRE-AMPLIFIER WITH A COMMERCIALLY
AVAILABLE INSTRUMENT
Detector: CdTe
Isotope : Co- 57
ORTEC Pre-Amp.
Mod. 142-A
(followed byOrtec
Research Amp.
Mod 450)
Pre-Amp. Cost:
8 495.00
S/N ~ 6.5
Time
SPECIALLY DE
SIGNED Pre-Amp.
(followed by Or tec
Research Amp
Mod 450)
Cost of Material:
S 25.00
S/N ~6.3
cont. on next page wm 012078
Figure 1.1
Amplifier Signal-To-Noise Comparisons


22
MAUDERLI Woller
429-72-4216
2 POINT SOURCES
ANGER TYPE
CAMERA
LEAD HOLES
S : Scatter orig. in
detector
R : Scatter orig. in
collimator
TUNGSTEN
PLATES
DETECTOR
lOverlapping sector of primary photon beam
reducing resolution
-reducing resolution*
PARALLEL
PLATE TYPE
CAMERA
DETECTOR
None of the [
above Problems 1
* Pulse-height discrimination does not eliminate scatter radiation when
the scatter angle is less than the following values: In case of Tc-99m
Lower discriminator level 5% below photon energy: ~ 50
10% : ~ 70
20% :~I00
(for Compton
scatter)
WM 012578
Figure 2 Detector-Collimator Arrangements


28
Figure 5 CdTe Detector (2x2x2 mm approximately)


71
full width at half maximum is essentially the width of one matrix
element, and the resolving capabilities of the camera are rather easily
i 11ustrated.
The performance of the camera was compared to the rotational
parallel plate imaging device (PPRID) designed and constructed at the
University of Florida (Luthmann, 1978). Although the ultimate sensitivity
possible with either device has not been determined, it is possible to
establish that the PPRID has superior sensitivity to a conventional
Anger camera (see figure 26) since it can resolve two parallel lines of
75 yCi each in 9.4 sec. As shown in figure 24, the camera developed in
this research can resolve two lines of 138 yCi and 68 yCi in only 3.8
sec. Although the ability of both cameras to resolve two parallel lines
appears to be about the same, the resolution of the camera described
herein is somewhat superior in that the transition from the no-activity
regions to full-activity regions is absolute (i.e., goes immediately
and directly from the lightest gray scale to the darkest gray scale).
The PPRID camera produces a multitude of gray levels around and between
the two lines (not evident in figure 26 due to photographic technique),
such that the full width at half maximum encompasses more than the width
of one matrix element. Although both cameras seem to have about the
same ability to resolve parallel lines, thyroid phantom cold spots, etc.,
in all cases the image is somewhat superior for the camera described
herein since the contrast is superior. This would be expected since
the PPRID depends upon a computed reconstruction of the image from a
multitude of related images. The PPRID camera also has positioning
errors associated with it that do not have a direct analogy in this
camera. The radiologic image is directly measured by the X Y CdTe camera
described herein and hence should produce better images.


76
the ability to use several collimator designs (as is done
on the Anger camera) without sacrificing the absorption of
scatter radiation between detectors.
6) The analog portion of the circuitry could be redesigned to
make the voltage output independent of the position of charge
deposition (and hence increase the proportion of detections
indicative of full energy photon detection). By utilizing
a circuit time constant that is much much shorter than the
collection time for electrons, the slope of the output
pulse (dV/dt) would be proportional to the number of ion
pairs produced and the magnitude of the output becomes:
V = R(dq/dt).
The output voltage pulse is proportional to the current flow
and hence its height is a measure of the number of ion pairs
created (Price, 1964). The closed loop gain of an amplifier
with such a short circuit time constant would be very low and
hence a second amplifier stage might be required. Design
considerations for low noise second stage amplifiers are
available (Arbel, 1968). Alternately, a precision microvolt
comparator could be used. The use of a very short circuit
time constant would facilitate higher sensitivity (higher
count rate) since pulse duration would be shorter. The
performance of the detectors and the performance of the as
sociated analog circuitry could be improved by cryogenic
operation (Elad, 1969).
7) Means for adjusting the timing of each analog to digital
electronics circuit would be useful in order to synchronize
the response to a given input pulse.


i
57
Exposure Conditions:
Source: Tc-99m point source (1mm diameter)
Geometry: close to collimator
Scales: arbitrary and unequal
+
XY CdTe Cameba
-POINT CENTERED
OVER ONE DETECTOR
Anger Camera
-HIGH RESOLUTION
COLLIMATOR (1A0 KEV)
Figure 16 Point Source Images


12
charge collection occurs over some time period. For charge liberation
near the negative electrode, high mobility elecrons traverse the
entire detector thickness and the low mobility holes traverse a much
shorter distance; hence charge collection requires less time than
those events in which holes must traverse longer distances. Ideally
the CdTe detectors should be irradiated with a normal incidence on the
negative electrode.
The approximate physical dimensions of the detectors available for
this research were 2x2x7 mm. The electrodes were two opposed sides
of 2 x 7 mm dimensions. The linear attenuation coefficient of CdTe
is approximately 0.36/mm for 140 keV photons and hence a 2 mm thick
detector should attenuate 60 percent of incident 140 keV photons (Hoffer,
1971). Photons of 140 keV are considered throughout this dissertation
since they are produced in the decay of technetium-99m (Tc-99m) which
is widely used in radioisotope imaging studies. The 7 mm detectors
will be cut into 3 detectors of approximate dimensions 2x2x2 mm.
The 60 percent attenuation of the photons of interest is sufficient
to demonstrate the characteristics of the X V CdTe camera.
The stability of the detector characteristics with time is important
since polarization phenomenon (i.e., changes in the internal electric
field with time) have frequently been found in CdTe detectors. It has
been reported that some p type, chlorine doped CdTe detectors with
platinum contacts are totally stable with time, do not exhibit polariza
tion and are sensitive throughout their volume (Serreze, 1974).
The bias voltage applied to the detectors will determine the
mean free path of the electrons and holes. To maximize charge collection,


53
The program also divides the image array into sixteen levels of
gray (coded 0 through 15) and formats the data in random access memory
for eventual use by a videographics interface. The sixteen gray levels
are derived by defining sixteen levels between 0 counts and the maximum
number of corrected counts per channel and assigning each matrix element
a value of 0 through 15 based on its corrected value.
Videographics Interface
The "Dazzler" videographics interface, manufactured by Cromemo
Incorporated, was utilized since it is the only interface readily avail
able that will read the random access memory of a S-100 compatible device
and display the data (properly formatted) in a 64 x 64 matrix with 16
shades of gray at each matrix location. The Dazzler does not require
servicing by the microprocessor during operation.
The Dazzler requires only 2048 bytes of computer memory. The
computer memory is scanned using direct memory access. The data in the
computer memory are formatted onto a TV screen to give a 64 by 64 matrix
with 16 shades of gray available at each matrix location. The Dazzler
output is a standard composite video signal that goes directly to a
monitor or television video amplifier. The microprocessor does encounter
some wait states caused by the Dazzler (while it reads the memory),
but this does not interfere with accurate program execution.
"In displaying the pictures, a compromise should be made between
contrast and gradation. Contrast should not be amplified far beyond
the point at which statistical variations become visible" (Anger, 1967,
p. 542). Gradation refers to the number of different gray tones used


67
Exposure Conditions:
Source: Tc-99m parallel line source 1mm diameter,
5,5mm apart)
Geometry: close to collimator
Scale: arbitrary
Activities: left line-68/^Ci
RIGHT LINE-13S/UtCI
Imaging Time: 5.8 seconds
,i
XY CdTe Camera
-LINES ALIGNED
WITH COLUMNS
-LINE LENGTH EXCEEDS
SENSITIVE VOLUME
Figure 24 Parallel Line Source Image (3.8 Seconds)


CHAPTER 2
HISTORICAL DEVELOPMENT
The radioisotope camera developed in this research is a unique
new design that has no direct predecessors. Some of the concepts utilized
in the camera have been tried in other types of radioisotope cameras;
however, the camera developed in this research is the first to employ
CdTe semiconductor detectors in anXY matrix and the first to locate
the detectors within the collimating holes. The research was original
in that a review of pertinent literature revealed the following:
1) Cadmium telluride detectors have not previously been
employed in an X Y matrix gamma camera radioisotope
imaging system. The idea to utilize CdTe in a radio
isotope imaging device is not original to this research.
Work is under way at several laboratories in regard to
the development of radioisotope cameras utilizing semi
conductor detectors (Entine, 1977). The use of CdTe
detectors in an X Y matrix, thus reducing the number of
outputs, has been offered for consideration (Allemand,
1977). One rudimentary imaging system utilizing an array
of CdTe detectors behind a parallel hole collimator was
developed (Zanio, 1977). The system utilized (n) outputs
(for an n x n matrix) and a nonuniform electric field
geometry. The detectors were joined electrically in an
attempt to fabricate a large volume detector. Although
the system functioned, it was not designed as a radioisotope
3


CHAPTER 7
CLINICAL FEASIBILITY
A clinically significant device should have a large enough sensitive
area to perform large organ studies or be large enough to be scanned or
composited to assimilate large organ images. Anger cameras have sensitive
2
areas as large as approximately 1000 cm Some of the considerations
for a clinical device are discussed below.
1) The matrix of a clinical device could be composed of sub
groups. This is dependent upon the number (or volume) of
CdTe detectors that can be operated in a row without
serious degradation of signal to noise ratio or pulse
height resolution. During testing of the small crystals
after cutting operations, it was demonstrated that at least
five of the small crystals could be operated in a row.or column
while still maintaining a distinguishable photopeak. The
manufacturer of CdTe has demonstrated that four to eight
detectors for a total aggregate volume of approximately
2 x 2 x 50 mm can be operated and still maintain a dis
tinguishable photopeak. The photopeak resolution of the
aggregate is substantially worse than any of the individual
detectors due to statistical signal to noise degradation
and due to the fact that no two detectors have essentially
the same photopeak pulse height which requires that the
detectors be roughly grouped according to pulse height
74