Light conversion, S/N characteristics of x-ray phosphor screens
Item Type text; Thesis-Reproduction (electronic)
Authors Lum, Byron Kwai Chinn
Publisher The University of Arizona.
Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
Download date 28/09/2021 05:29:31
Link to Item http://hdl.handle.net/10150/557456 LIGHT CONVERSION, S/N CHARACTERISTICS
OF X-RAY PHOSPHOR SCREENS
by
Byron Kwai Chinn Lum
A Thesis Submitted To the Committee on
COMMITTEE ON OPTICAL SCIENCES (GRADUATE)
In Partial Fulfillment of the Requirements for the Degree of
MASTER OF SCIENCE
In the Graduate College
THE UNIVERSITY OF ARIZONA
19 8 0 STATEMENT BY AUTHOR
This thesis has been submitted in partial fulfillment of re quirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.
Brief quotations from this thesis are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his judg ment the proposed use of the material is in the interests of scholar ship. In all other instances, however, permission must be obtained from the author. ,
SIGNED: 'K.C. / -
APPROVAL BY THESIS DIRECTOR
This thesis has been approved on the date shown below:
OltiUV 4- HANS ROEHRIG I Date Adjunct Associate Professor of Radiology ACKNOWLEDGMENTS
I would like to express my deep appreciation to my advisor.
Dr. Hans Roehrig, whose guidance and patience made this thesis possible. I am also very grateful for the efforts of Ms. Betty
Porter and Ms. Delia Bryant in the preparation of the final copy of this thesis.
This work was sponsored under the project "Evaluation of
PEID Systems for Radiology", awarded through the Bureau of
Radiological Health, Food, and Drug Administration under Grant
No. 5R01FD00804-04RAD. TABLE OF CONTENTS
Page
LIST OF ILLUSTRATIONS...... v
LIST OF TABLES ...... viii
■ A B S T R A C T ...... ix
1. INTRODUCTION ...... 1
Justification ...... 2 Physical Processes of X-ray Induced Emission ...... 9 Characteristic X-ray Reabsorption...... 14 Properties of New and Traditional X-ray Phosphors . . . 20
2 NOISE ...... 26
Introduction ...... 26 Statistics of Screen Amplification ...... 34 Scintillation D Q E ...... 38 Simulation of Screen Statistical Processes . 40 Some Conclusions ...... 44
3. EXPERIMENTAL PROCEDURE AND SE T U P ...... 45
Introduction ...... 45 The Photomultiplier ...... 47 The X-ray Source ...... 35 Electronics...... 36. Computer Analysis ...... 38
4. RESULTS AND CONCLUSIONS...... 39
R e s u l t s ...... 39 Analysis ...... 72 Conclusions ...... 79
REFERENCES...... 81
iv LIST OF ILLUSTRATIONS
Figure Page
1. Screen-film combination (Messier 1973) 3
2. Theoretical and experimental values for radiographic noise; Experimental-theoretical (Rossmann, 1962) ...... 6
3. PEID, using fluorescent screen optically coupled to intensified video camera tube ...... 7
4. Schematic of x-ray intensifier video camera system..... 7
5. Cross section for interaction in a calcium tungstate (CaWO^) screen (Vybomy, 1978) ...... 11
6 . X-ray absorption processes in a Csl phosphor and the resulting absorbed energy spectrum for monochromatic x-rays (Swank, 1973) ...... 12
7. Inner-electron transitions for characteristic x-ray emission (Weidner and Sells, 1973) ...... 15
8. Characteristic x-ray spectrum for a thulium (Tm) secondary t a r g e t ...... 15
9. X-ray attenuation as a function of energy for a Ba phosphor (Vybomy, 1 9 7 8 ) ...... 16
10. Characteristic x-ray re absorption diagram (Vybomy, 1978) . . 16
11. CsBr:(Tl)1, CsI:Na(2), ZnCdS:Ag(3), and CsI:Tl(4) ...... 22
12. BaS04 :Eu2+ (5), BaFCl:Eu2+ (6), and CaW04(7) ...... 22
13. Gd„(LS:Tb(8) , La„CLS:Tb(9), and Y 0„S :Tb(10) . (Stevens, 1975) 7 7 ...... 23
14. Radiographic mottle (Shaw, 1976) 27
v vi
LIST OF ILLUSTRATIONS--Continued
Figure Page
15. Components of radiographic density fluctuations (Rossmann, 1962) . 27.
16. Spatial Frequency content of noise in radiographic images (Rossmann, 1962)...... 29
17. Random fluctuations in density (SPSE Handbook of Photographic Science and Engineering, 1973)...... 30
18. Processes in the variation of x-ray screen scintillations. . 32
19. Schematic for the serial combination of two statistical devices (RCA Photomultiplier Handbook, 1970) ...... 34
20. Simulation screen probability distributions ...... 41
21. Simulation results for Fig. 20(a), P^ - P^ = 0.5 ...... 42
22. Simulation results for Fig. 20(a), P^ - 1/4, P^ = 3/4 . . . 42
23. Simulation results for Fig. 20 . = . 43
24. Block Diagram of U of A evaluation facility ...... 46
25. Schematic of PMT pulse counting method ...... 46
26. RCA 8850 photomultiplier pulse height spectrum ...... 48
27. Photomultiplier output corresponding to a Poisson distribution ...... 49
28. Phosphor screen output decay characteristics ...... 50
29. Decay time constant for a CaWO^ screen ...... 50
30. Calibration of photomultiplier counting efficiency ...... 51
31. Counting efficiency of the system ..... 53
32. Variable energy x-ray source ...... 55
33. System electronics ...... 57 vii
LIST OF ILLUSTRATIONS--Continued
Figure Page
34. Measured probability distributions (P ) for a ZnCdS screens: Cd K-edge: 26.7 keV ...... f ...... 60
35. Measured probability distributions (P ) for a CaWO. screen; W k-edge = 69.5 k e V ...... f ...... 61
36. Measured probability distributions (P ) for a BaSO. screen; Ba K-edge: 37.5 k e V ...... 7 ...... 62
37. Measured probability distributions (P ) for a La.O S-Gd^O S screen; La k-edge: 38.9 keV Gd k-edge: 50.2 k e V ...... 63
38. Measured probability distributions for a Csl x-ray image intensifier; 1 k-edge: 33.2 keV, Cs k-edge: 35.9 keV .... 64
39. Bremstrahlung spectrums for different amounts of filtrations ...... 65
■40. Measured output emissions for a CaWO^ screen ...... 66
41. Measured output emissions for a BaSO^ screen ...... 67
42. Measured output emissions for a series of ZnCdS screens . . . 6 8
43. Measured output parameters for a La^OgS-GdgO^S screen...... 69
44. Measured output parameters for a Csl x-ray image intensifier ...... 70
45. Measured signal to noise ratios from the output of Csl x-ray image intensifier experimental values below 10 absorbed x-ray photons in error due to the fact that the threshold counter did not sample continuously, but was triggered by signal pulses ...... 73
46. Measured signal to noise ratios from the output of a CaWO^ screen; 44 keV incident x-ray photons ...... 74
47. Csl image intensifier output distributions ...... 77
48. CaWO^ screen output distributions .....'...... 78 LIST OF TABLES
Table Page
1. Probability of barium k reabsorption in a pair of barium strontium sulfate screens versus incident x-ray energy .... 19
2. Reabsorption probabilities forthe K x-rays emitted by the principla phosphor elements in the screens studied ...... 19
3. Basic properties of phosphors ...... 21
4. X-ray source characteristics ...... 53
5. Measured output efficiencies of some phosphor screens .... 71
viii ABSTRACT
The variations in gain or amplification are measured for a variety of x-ray phosphor screens and for a Csl image intensifier as a function of incident x-ray energy. These variations result in a reduction of the output SNR (signal to noise ratio) by a factor of
/DQEsc^nt• The scintillation detective quantum efficiency, DQ^scint' is evaluated theoretically and experimental results are presented.
The findings show that the newer rare earth phosphor screens possess a higher gain than do the traditional calcium tungstate (CaWO^) screens and that the values for DQEsc^n t , do not vary considerably for a different phosphor materials. CHAPTER 1
INTRODUCTION
Since the incident x-ray photons follow Poisson statistics, the input SNR (signal to noise ratio) of an x-ray imaging system is readily known. However, for systems in which x-ray intensifying phosphor screens are utilized, the output SNR following screen amplification will be degraded due to variations in the amplification or gain of the phosphor screen. In this thesis, the average gain and the associated variations for various phosphor materials are measured and its effect on the SNR is investigated theoretically and experimentally. The experimental measurements are done with a photon counting system with which individual x-ray absorption events may be detected and analyzed.
The remainder of this chapter will concern itself with the justification for this study and with the basic concepts of x-ray induced fluorescence. Also included at the end of the chapter is a brief run-down of the various phosphor materials that are, or will be available for diagnostic radiology. Chapter 2 deals with the theo retical analysis of the noise that is associated with x-ray intensify ing screens and includes a simulation study of the screen's statistical processes. The term "scintillation detective quantum efficiency,"
DQEscint’ introduced and formally defined. In Chapter 3, a detailed description of the photon counting system is given, where the principal components are the x-ray source, photomultiplier, electronics, and the digital computer. Finally, the experimental results are presented in Chapter 4 and comparisons are made between the theory and these results.
Justification
For the past fifty years medical radiology has consisted of the transmission of ah x-ray beam through the patient and the recording of this beam on a screen-film combination. A screen-film combination is basically a sheet of film sandwiched between two fluorescent intensi fying screens (Fig. 1). The phosphor in the screens has been tra ditionally CaWO^, which emits UV and visible light photons with the absorption of a single x-ray photon. The phosphor screens thus pro vides a gain mechanism in the process. Quite naturally, in the interest of the reduction of patient dosage, one would want the screens to possess good x-ray absorption characteristics and high light output. Recent years have seen the advent of new rare-earth screens which have relatively higher x-ray absorption capabilities and better x-ray to light photon conversion ratios.
However, a higher signal does not necessarily imply a better image. What are also important are the signal to noise relation ship and the spatial resolution capability. For instance, if the larger signal is characterized with even larger fluctuations, and if detail in the image is smoothened out as a result of this gain in signal, then the effort would seem hardly worthwhile. Past papers 3
Screen Base
deflective Coating X-Ray Excited Phosphor
Protective Coating
‘X-Ray Film Base
lulsion
W W W J M i M M
Fig. 1. Screen-film combination (Messier 1973).
(Cleare et al., 1962; Rossmann, 1962; Rao and Fatouras, 1979) have given some considerations to the noise aspect of phosphor screens, but these studies seem to totally ignore the physical characteristics of the intensifying screens. For instance, they do not consider the various interactions that may occur between the incident x-ray photons and the screen material. These interactions may cause variations in the scintillations of a screen and thus affect the noise characteristics.
One paper of particular interest (Rossmann, 1962) deals with the comparison of theoretically and experimentally determined noise characteristics of a particular x-ray radiograph. In the derivation of the theoretical expression, Rossmann first assumes that the ex posure E determined by the number of x-rays absorbed by the screen
in a given area in a given time is n^a, where a is the film area of
interest and n is the number absorbed per unit area. If the fluc- x tnations in this number is Poisson limited, then the standard deviation would be given by:
ox = ^ - i C D
In order to relate this quantity to the density fluctuations in the film, one must consider the characteristic curve of the film derived from a plot of the density D versus the logarithim of the exposure
E, i.e.
D = Y log10E + C (2) where y is called the film gamma and is the gradient of the H and
D curve and C is the density due to fog and base. Differentiating, we obtain
AE ' 0x AD = 0.43 y — - . = 0.43y --- - E n a (3) x
Through substitution and taking into account film granularity, the total density fluctuation is
(0.43a)2'^ a(D)total = {a (D)grain + ------^ (4) nxa
What now remains is to take into consideration the modulation transfer function (MTF) of the screen-film combination and of the scanning aper ture. If we assume that the scanning aperture is circular, its MTF is given by where d is the diameter of the aperture and is a first order Bessel
function. Denoting the MTF of the screen-film combination as A (v), the measured Wiener spectrum of the mottle is given by:
WtleM = |A#(v )|2 (6) where K is a constant. Assuming that the MTF of the screen-film system
is of the form
A#(v) = exp(-27r^p^v) (7) and since
o 2 (D) = ^mottle^Vx^Vy (8) we have as the total density fluctuation
2 (0.43y)2F2 , total - {" (D1 grain + ( ’ nxa where Fisa function that varies between zero and one. Figure 2 shows the comparison between the calculated fluctuations and the measured
fluctuations. The film was Kodak Blue Brand Medical x-ray film with
Kodak fine-grain screens. The tube voltage was 80 KVp with a % mm Cu
filter at the tube.
It is clear that the correlation between theory and experiment
is quite good, and one can conclude, therefore, that the noise is essentially Poisson limited, assuming the film granularity is negli
gible. . Note that no considerations were made of the screen character
istics nor of the beam quality. A recent study done by Rao 6
.01
e> .005
800 1200 d (microns) Fig. 2. Theoretical and experimental values for radiographic noise; Experimental-Theoretical (Rossmann, 1962).
and Fatouras (1979) reveals, however, that there are some notable
discrepancies between theoretical and experimental values, contrary
to the conclusions. These results suggest that other secondary
sources of noise, such as gain variations with the phosphor screen,
may contribute to the final noise output.
Of recent interest to diagnostic radiology has been the
application of photoelectronic imaging devices (Nudelman, 1976,
Beckmann, 1978). In order to evaluate the performance of these
devices, in particular the signal to noise ratio, one has to con
sider the gain and losses of the signal as it works its way through
the system. For instance. Figure 3 shows an intensifying screen
optically coupled directly to a video camera tube. In Figure 4,
the output from an x-ray image intensifier is optically coupled to the camera. In calculating the signal incident upon the camera,
for both cases, one takes into account the gain of the image CRT DISPLAY IMTfHSIFICD CAMERA TUUE X-RAY AMPLIFIER SOURCE
Fig. 3. PEID, using fluorescent screen optically coupled to intensified video camera tube.
m: i Ki n i.A ion
Cs_Sb Re P720
«A»
t IS AVI: MAK
Mill. IHINIiASI; NISISIIW lAI'ACITAW.’ti
UISINIINIIHIN SNR = / n .P.m2A .tcD(^Escint
Fig. 4. Schematic of x-ray intensifier video camera system. intensifier, the transmission of the optics, and the numerical aperture of the optics. If the variations in the gain of the image intensifier is taken into account, then the SNR at the output of the video camera
[Fig. 4) can be shown to be approximately:
2 DQE h (SNR)VID “ (NxPxM Vf ' ----— ) (10) 2 where = x-ray photon flux (photons/cm - sec)
n^ = x=ray image intensifier quantum absorption efficiency
m = overall system magnification
Ag = video camera tube pixel area
tj. = video camera tube frame time
a = factor associated with secondary electron emission
due to the landing beam
DQEscint = scintillation detective quantum efficiency
The quantity DQE^^^. will be defined in the following chapter, but for the present, it would suffice to say that it is a measure of the vari ation in the intensifier1s gain. Furthermore, if the collection effi ciency of the optics is very low, then the gain of the intensifier or of the screen must be high enough so that there may be a detectable signal for the camera. This same principle can be applied to a simple radio- graphic screen-film system, in that the gain of the intensifying screens must be high enough to account for the low quantum efficiency of the film.
Presently, much attention has been given to the subject of human utilization of the information present in the radiographic image (Wagner, 1977). Quantitative characteristics of the imaging system such as resolution, gain, noise, etc. are needed to be incorporated with the human-eye brain system in order that the human response to the final image may be predicted.
From the above discussion, it is not difficult to realize the importance in the evaluation of radiographic imaging systems. A system that may actually "count" the number of light photons emitted by a particular device would be highly desirable in achieving this end.
It would be externely useful for the determination, of the gain of an x-ray image intensifier or of a specific phosphor screen, notably the
"brighter" rare-earth screens as compared to the CaWO^ screens. A photon counting technique is also very desirable for a signal to noise evaluation since it permits an accurate measurement of signal fluc tuations in terms of photon quanta. This method is highly suitable for Poisson processes which deal with discrete variables.
Physical Processes of X-Ray Induced Emission
The gain of an x-ray phosphor screen consists of the conversion of a single highly energetic x-ray photon into several hundred to sev eral thousand less energetic light photons. Thus, in the case of a screen-film system, many silver halide grains are made developable in comparison to only one or two if the x-ray photon was to be absorbed directly by the film. The phosphor screen also offers two advantages over film; its ingredients offer a larger mass absorption coefficient that silver and it is thicker than film, thus providing better 10 absorption efficiency. These two factors of phosphor gain and better quantum efficiency combined can result in a gain of approximately 50 when considering a double film-screen system.
For the standard medical x-ray energy ranges, an x-ray beam is attenuated by the phosphor screen through three interactions; 1) photo electric absorption, 2) Compton or incoherent scatter interaction, and
3) Rayleigh or coherent scatter interaction. The cross sections for these processes are shown in Figure 5, indicating that the dominant interaction is the photoelectric effect. Intensifying screens are usually comprised of inorganic phosphor crystals that contain elements with high atomic numbers. As a result, an x-ray absorption causes the ejection of an electron from the K or L shell of the host atom.
The kinetic energy of the electron is the difference between the absorbed photon energy and the binding energy of the atom. These energetic electrons> through inelastic collisions, ionize other host atoms, thus producing secondary electron-hole pairs. In addition, the initially excited host atom, which now has K or L shell vacancies, will eventually relax, resulting in the emission of secondary, less energetic, x-ray photons or Auger electrons. These may then be absorbed by the phosphor medium, thus contributing to the electron- hole pair production. In the final step of the process, the second ary electrons excite still other host atoms or activator impurities into higher energy states. These excited atoms eventually decay, resulting in light fluorescence. 11
CaWO
10*
^worocLEcrmc
20 30 *0 30 60 so 90 too photon energy (keV)
Fig. 5. Cross section for interaction in a calcium tungstate (CaWO ) screen (Vyborny, 1978).
The energy conversion efficiency, as one would expect, is by no
means 100%. As mentioned above, the atom with a K-shell vacancy will produce a characteristic K-shell x-ray or an Auger electron. The
Auger electron is most often reabsorbed, but the K x-ray has a
relatively high probability of escape from the screen and thus does not
contribute to the fluorescence process. This particular facet will be
discussed in further detail later since it also plays a major role
in the detective quantum efficiency of phosphor screens. A diagram
illustrating these various processes is shown in Figure 6 . 12
INCIDENT X-RAYS
LM... LK. . . SHELL SHELLS SHELL SHELLS
CD
<
N(E)
Fig. 6 . X-ray absorption processes in a Csl phosphor and the resulting absorbed energy spectrum for monochromatic x-rays (Swank, 1973). 13
Another mechanism that can result in lost energy involves the production of photons. Present literature (Kingsley, 1975) states that lattice vibrations contribute very little to energy loss until the secondary electron-hole pairs have energies of about 10 to 15 eV.
A generally accepted empirical rule states that it takes on the average an amount of energy that is three times the bandgap energy in order for a single electron-hole pair to be created. As a result,
2/3 of the available energy is converted into heat (Kingsley, 1975).
Three other sources contribute to reduced energy conversion.
First an electron emission reduces the efficiency of electron-hole pair production, and secondly, the final luminescence process has an efficiency of about 90%. The third source stems from the fact that the photons must travel through the screen before it can be emitted.
Approximately half of the photons can be lost in this final process.
The above processes can be exemplified by considering the case of a zinc sulfide (ZnCdS) screen, which has a bindgap energy of 3.3 eV.
Taking into account losses through lattice vibrations, it would require approximately 10 eV to create a single electron-hole pair. If a 60 keV x-ray photon is absorbed, then about 6000 electron-hole pairs are pro duced. Since the luminescence efficiency is about 90%, this would result in about 5400 light photons. If the impurity activator is silver, then the emitted photons would have an energy of = 2.8 eV, which is in the blue region of the visible spectrum. The energy of conversion efficiency is therefore in the order of: 5400 x 2.8 eV _ _ 0 60 keV
The highest intrinsic efficiency found for any phosphor is in the neighborhood of this value (Kingsley, 1975).
Characteristic X-Ray Reabsorption
When the absorption of an x-ray photon occurs above the k-edge of the phosphor element, approximately 90% of the excited phosphor atoms decay radiatively (Vybomy et al., 1978). This radiative decay results in the production of Ky and Kg x-rays whose energies are lower than that of the incident x-ray energy. This process is illustrated in Figure 7, with the resultant characteristic spectrum in Figure 8.
Since the energies of the k x-rays are slightly below the k-edge of the phosphor element, as illustrated in Figure 9 for the case of a Ba phosphor, many may escape the screen entirely due to the low absorption coefficient.
Theoretical calculations for the probability of k-escape have been done by Vybomy et al. (1978) and the analytical derivations are shown in the following discussion. We first divide the screen into
N slices and the total solid angle subtended from the center of the slice into R pieces. For a given nth slice as shown in Figure 10, the average mean free path for a characteristic x-ray emitted within the rth solid angle is: W 15 0 N
M
L
K
Fig. 7. Inner-electron transitions for characteristic x-ray emission (Weidner and Sells, 1973).
Counts
J
49.8 50.7 57 5 X-Ray Energy (keV)
Fig. 8. Characteristic x-ray spectrum for a thulium (Tm) sec ondary target. 16
1.00 .90 .80 .70
.10 20 3 0 4 0 50 60 70 80 9 0 1 0 0 PHOTON ENERGY (keV)
Fig. 9. X-ray attenuation as a function of energy for a Ba phosphor (Vyborny, 1978).
FRONT SCREEN
Fig. 10. Characteristic x-ray reabsorption diagram (Vyborny, 1978) where cose^ is the average value of cos8^ within this particular solid
angle, expressed as \ 2 cos0sin6d0 3rl cose r (12) 5r2 sinOde -°rl
Assuming that scatter is of no consequence in the reabsorption process, then the characteristic absorption probability P will be a n , r given by
Pa,n,r = {1"exP I " (Pa/p) (N-y Wp/Ncos0r | } (13)
where is the absorption coefficient of the material for an x-ray
energy of and p is the density of the screen material. The average
probability for the entire slice is then obtained through summing
over the R solid angles and dividing by R,
R p = _ i y p (14) a,n R ^ a,n,r r=l
These probabilities must now be weighted by the probability ,
which is the relative probability of having a k emission in the nth
slice for an incident x-ray energy of E^. is simply the number
of photons incident upon the nth slice multiplied by the total number
of absorbed in that particular slice and finally divided by the
total attenuation of the screen: 18
exp{- (n-1) | Wp((yT+y T) /p) |/N}{l-exp| -((y +y )/p)Wp/N | } G - ---- ;------— ------(15) {1-exp| - (( y I +y^I )/p)Wp | }
Note that the above expression takes into account incoherent and
coherent scatter. The total characteristic probability of k-reabsorp-
tion is thus given by:
N
p T = y g Tp al L . nl an n=l
N (16)
P6I = L l^ S "
V y b o m y has carried out the calculations for various phosphor screens,
including ones that contain two principal absorption edges. Their
results are shown in Tables 1 and 2. One conclusion that can be made
is that the reabsorption probability is independent of incident x-ray
energy. Secondly, the probabilities range in values from - 0.2 to
0.6, indicating that a significant portion of the incident energy is
lost. Besides reducing the energy conversion efficiency, this partic
ular phenomenon also plays a large role in the signal to noise
considerations of x-ray screen systems. This topic will be covered
in more detail in the following chapter. However, since many of the
interactions involve the production of k x-rays, a large portion of
the energy absorbed by the screen is through k-reabsorption whenever
the incident energy is above the k-edge of the phosphor. \ 19
Table 1. Probability of barium K reabsorption in a pair of Barium Strontium Sulfate screens versus incident x-ray energy.
Incident x-ray energy, keV X-Omatic Regular
37.4 0.495 40 0.497 50 0.500 60 0.500 80 0.501 100 0.501
Table 2. Reabsorption probabilities for the K x-rays emitted by the principal Phosphor elements in the screens studied.
Screen Pal P61 Pa2 V
Par 0.21 0.16 Hi-Plus 0.32 0.24 Lightning-Plus 0.40 ,0.32 X-Omatic Regular 0.50 0.40 Alpha-4 0.48 0.37 0.40 0.32 Lanex-Regular 0.64 0.54 0.53 0.44 20
Properties of New and Traditional X-Ray Phosphors
In recent years, research on phosphor screen development has brought forth a large number of new rare-earth phosphors whose characteristics appear to be an improvement over the traditional
CaWO^ and ZnCdS phosphors. Table 3 lists the basic properties of these phosphors, while Figures 11-13 show their emission spectra.
The term "luminescent radiant efficiency" is simply the ratio of the energy emitted by the screen to the energy absorbed by the screen. One can immediately see that the k-edges of these materials are at lower energies than the tungsten k-edge. This latter characteristic can be shown to result in higher absorption coeffi cients for the range of x-ray energies' in diagnostic radiology.
Also note that the emission spectra of some of the newly developed screens are different from that of CaWO^, being shifted towards the green portion of the spectrum. This may pose some problems in terms of detector adaption. The following paragraphs will provide some brief comments on each of the above phosphors.
CaWO^ (Coltman, 1947) is the traditional phosphor used in x-ray intensifying screens and has well established manufacturing procedures and properties. The other traditional phosphor, ZnCdS: Ag
(Ludwig and Prener, 1972) is used mainly as a direct viewing (fluoro- scope application) screen and as the input and output phosphor of image intensifiers. Table 3. Basic properties of phosphors
* PHOSPHOR Z n**
BaFCL:Eu2+ 56 13 BaSo^:Eu2+ 56 6 CaW04 74 3.5 CsBr:T1 35/55 8 Csl:Na 53/55 10 Csl:T1 53/55 11 Gd202S:Tb 64 15 La202S:Tb 57 12 Y202S:Tb 39 18 ZnCdS:Ag 30/40 18
*Z = atomic no. of principal absorber **n = luminescent radiant efficiency (%) 22
100
80
60
40
f » Jo 300 400 500 600 700 A(nm)———♦
Fig. 11. CsBr:(Tl) 1, Csl:Na(2), ZnCdS:Ag(3), and Csl:T1(4).
oo
80
40
300 350 400 450 500
Fig. 12. BaS04 :Eu2+(5), BaFCl:Eu2+(6), and CaW04(7). 23
too 8 r 5 o lii. . 400 500 600 700 A(nm) ►
100 9
♦ 50
— ... \ ,I i 1 400 500 600 700 A(nm)— »
XX)
50
500 600 700
Fig. 13. Gd O S:Tb(8), La O S:Tb(9), and Y 0 2S:Tb(10) (Sfevens, 1975). 24 2 BaFCLrEu is being considered as an intensifying screen due to its higher conversion efficiency and absorption capabilities. It has 2 2 however an "afterglow" drawback. BaSO^Eu is similar to BaFCL:Eu , except it has a lower conversion efficiency. Both of these screens possess the advantage of having their emission spectra towards the blue, which is compatible with the traditional blue-sensitive films
(Messier and Wolfe).
Cdl: Na and Csl: T1 have been used in replacing ZnCdS: Ag as the input phosphor of image intensifier tubes (Bates, 1969), due mainly to the element's higher mass absorption coefficients and density. This property is very important since the input phosphor thickness of intensifiers are typically on the order of only 10 mils (Ludwig and Prener, 1972).
Gd202S:Tb (Buchanan, 1972) is a screen being proposed as the input screen for image intensifiers, as an intensifying screen in filmscreen combinations, and as a direct viewing screen. However, the densities of Gd202:Tb screens are very low and thus possess the same disadvantage as ZnCdS:Ag screens, Its other disadvantage stems from its emission spectrum which is at around 550-600 nm, making the coupling to traditional blue sensitive films difficult. La.202S:Tb has the prospect of being an intensifying screen and was at one time a contender for the input screen of x-ray intensifiers. Its emission spectrum is similar to that of Gd202:Tb. Two other practical screens however, which have emission spectra in the blue are Gd202Br and La.202Br. Despite its relatively high conversion efficiency, YgOgSzTb is a rather poor absorber when compared to Gd^OgSzTb and La^OgBr.
However, its emission can be shifted from green to blue, and thus may be used with standard x-ray film. For these reasons, ^C^StTb has the potential of becoming an intensifying screen for a film-screen combination (Alves, Buchanan, 1973). CHAPTER 2
NOISE
Introduction
If one were to examine a uniformly exposed radiographic image, it would be fairly easy to notice the granularity or mottle present (Fig. 14). A first logical choice as for the cause of these density fluctuations might be film granularity, but experiments show that this plays a relatively minor role (Cleare et al., 1962).
Images that were obtained with varying distances between the screen and the film show that the mottle pattern changes dramatically. Only when the screen-film distance is very large do the high spatial frequency fluctuations that are associated with film granularity become dominant. A second choice might be the structural inhomo geneities in the phosphor coating of the screens, which is appro priately termed "structure mottle." Two identical successive x-ray exposures were made on the same area of an intensifying screen in hopes of finding some correlation between the two images. However, no such correlation was ever found, suggesting that structure mottle also plays a very minor role.
In view of the above findings, one might safely conclude that the density fluctuations of the film are primarily associated with the statistical distribution of x-ray photons that are absorbed by
26 27
Fig. 14. Radiographic mottle (Shaw, 1976). the screen. This third component of the radiographic mottle is referred to as "quantum mottle." Quantum mottle easily explains why the mottled appearance decreases as one goes from a fast film to a slow film.
The slower film would require more absorbed x-ray photons in order to achieve the desired density, thus increasing the signal to noise ratio in the image. Similarly, screens with higher intrinsic con version efficiencies would theoretically increase radiographic mottle.
Figure 15 is an illustration of the above discussion.
Radiographic mottle
Screen mottle
Quantum mottle Structure mot 11e grain!ness
Fig. 15. Components of radiographic density fluctuations (Rossmann, 1962). 28
However, in terms of the "appearance" of the mottle, purely
statistical considerations cannot account for the results of some
radiographic images. For instance, it was mentioned earlier that the mottle is "smoothened out" as the screen-film distance is increased, whereas the quantum mottle explanation would have given identical
images regardless of the distance. This observation suggests that mottle is highly dependent upon the optical properties of the imaging
system. This is very important in view of the fact that the screen-
film combination isn't simply an amplifier, but also an imperfect
imaging system. This concept is rather obvious when we consider the
input to be Poisson limited (i.e. white noise) and that some of this noise will be eliminated due to the limited spatial bandwidth of the system. As derived in the section on the justification, the total
density fluctuation of a film-screen combination is given by:
(0.43Y)2F2 ^ “(“hotal - ™grain * — =------n a x The fluctuation F carries the information about the system's modulation
transfer function (MTF), where
0 poor MTF F = (17) 1 perfect imaging
Figure 16 shows the spatial frequency content of the noise typically
found in radiographic images. The above expression clearly explains
the behavior of the mottle pattern as the imaging properties of the
system are varied. 29
perfect
i/i tota c.
l/)
O >
c
grain
v (cycles/mm) Fig. 16. Spatial frequency content of noise in radiographic images (Rossmann, 1962).
At this point of the discussion, it would be wise to formally define such terms as "noise" and "detective quantum efficiency."
Again, consider a sheet of radiographic film that had been uniformly exposed. As described previously, the developed film will contain random fluctuations in density as illustrated in Figure 17. Noise is defined as simply the measure of the random fluctuations about a mean signal. In this case, it would be equal to o^, implying a signal to noise ratio (SNR) of
SNR = Do / o d (18) 30
I>
Distance
Fig. 17. Random fluctuations in density (SPSE Handbook of Photographic Science and Engineering, 1972). where = average density. Detective quantum efficiency (DQE) compares the input SNR and the output SNR of a particular system or device. It is defined as:
(SNR)2out DQE = r (19) (SNR) in
A device that does not degrade its input SNR has a DQE of 100%.
The input signal to noise ratio of the system is defined as the square root of the number of x-ray photons that are absorbed by the intensifying screen. Note that the SNR_^ is determined by the number of absorbed photons, not incident photons. With this defini tion of an input SNR, the quantity "scintillation DQE" may now be defined. It simply represents the degradation in the SNR due solely 31
to the variation in the gain of the phosphor screen. It does not take into account the decrease in SNR due to the absorption of the
screen. If one were to take into account the absorption, then
the term "screen DQE" might be more appropriate. These two defi nitions are related as in Equation 20,
screen " A x ^scint C20) where A = absorption of the screen. This discussion and later evaluations will deal only with the "scintillation DQE".
It is not very obvious as to what happens to the signal to noise ratio after (phosphor screen) absorption and amplifications.
If one were to simply count the number of absorbed x-ray events in the screen, then the output signal to noise ratio should essentially be identical to that of the input. Unfortunately, most detectors are
integrators and consequently do not simply count photon events.
Due to the various processes that occur in the screen, it should be obvious that the signal output from each absorbed x-ray event will vary, and it is this variation that degrades the signal to noise
ratio.
The variation in the signal pulse from an individual absorbed
x-ray photon is due mainly to three factors: 1) the incident x-ray
energy distribution, 2) the screen’s energy absorption distribution
for a given monochromatic input energy, and 3) the screen output signal
distribution for a given amount of absorbed energy. The first factor .
arises from the fact that the incident x-ray energy distribution for 32 a standard radiographic x-ray generator is a Bremstrahlung distribution.
The second factor is related to the absorption properties of the screen, e.g. k-escape or k-reabsorption and phonon losses. Lastly, the third factor deals with the optical properties of the screen where the light pulse from a given x-ray photon absorption is attenuated as it prop agates through the screen before it is emitted. These three processes can be represented schematically as shown in Figure 18. The final pulse height distribution is given by the integral:
Ps(E) = | H(E,E1) {F(EM)G(E’,EM)dE”}dE' (21) where
F(E") = input energy spectrum
G(E',E") = absorption spectrum
H(E,E*) = optical pulse spectrum
Ps(E) = screen probability distribution
F(E") G(E',E")H(E,E ') ► Ps (E)
Fig. 18. Processes in the variation of x-ray screen scintillations.
Thus the signal from an absorbed x-ray photon, for all practical pur poses, will not be constant and will have a probability distribution as a function of output scintillation energy, P^(E). 33
In order to illustrate how P (E) affects the signal to noise
ratio, let us assume a hypothetical situation where we have a mono
chromatic x-ray source and that the variations in the optical absorp tion are negligible. Furthermore, assume that the screen has a k-edge at 50 keV and that there are 100 60 keV x-ray photons absorbed within a given area within a given time. Also, for the sake of clarity, let us suppose that the intrinsic conversion efficiency is identical for every absorbed x-ray photon. In the first example, it is assumed.that no energy is lost through k-fluorescence. Con
sequently, the total output signal will be proportional to 100 x 60
KeV = 6 MeV and the fluctuation in this signal will be proportional to 100 x 60 KeV = 0.6 MeV. The resultant signal to noise ratio
is therefore 6 MeV/0.6 MeV = 10, which is equal to the input signal to noise ratio.
Now consider the case in which k-escape occurs half of the time. Assuming that the k x-ray photon that escapes the screen has
an energy of 40 keV, the energy output would be proportional to
50 x 20 keV + 50 x 60 keV = 4.0 MeV and the fluctuation proportional to (/5(T x 60 keV) + (/SO" x 20 keV) = .566 MeV. The output signal to noise ratio is therefore 7.07, corresponding to a screen detective
quantum efficiency (DQE) sc£n1: of = 50%.
From the above examples, one can conclude therefore that the
signal to noise ratio degradation occurs not from energy losses, but
from the fact that the output signal per absorbed x-ray photon is not 34 constant. It is easy to see that the DOE . would still be 100% scint even in the case where k-escape occurs 100% of the time. Finally, if one were to allow also for variations in the intrinsic conversion efficiency for each absorption event, the output signal to noise ratio would be further degraded.
Statistics of Screen Amplification
A more formal approach towards analyzing the screen's contribution to noise can be made with the aid of Figure 19. Device
A would represent the input Poisson distribution for the absorbed
DEVICE DEVICE ->• nAB= V nB A "A B — 2 + n nA,aA v V r °AB '("o' ‘ o 2 " — A —
Fig. 19. Schematic for the serial combination of two statistical devices (RCA Photomultiplier Handbook, 1970). incident x-ray photons while Device B represents the screen's probability distribution P^(m). The various parameters are defined as follows:
n^ = average number of absorbed x-ray photons
within a given area and time
= Variance in the number of absorbed x-ray
photons = n^ 35
rig = average screen scintillation output per
absorbed x-ray photon 2 Og = variance in the screen's average scintil
lation output per absorbed x-ray photon.
Before continuing any further, some basic statistical principles ought to be reviewed. For a general probability function
P(n), where n is the number of particles for a given event, one has the basic property of:
N
I?(n) = 1 (22 ) n=o
With a given probability distribution, it is possible to create what is called its generating function, defined as
N Q(s) = Is11 P(n) , (23) n=o where s is an auxiliary variable. This particular function is useful
in that it has the following properties:
(24)
3Q(s) n (25) as s=i
92Q ( s ) (26) 3s2 s=l 36 Thus, given a generating function that describes the statistical
properties of a system's output, its mean, standard deviation, and probability distribution can, in theory, be derived.
Applying these concepts in evaluating the output characteristics
of an intensifying screen, the first step would be to find the generat
ing function of this system. For a serial combination of statistical processes, the generating function is given by
s Ws)} • C”) where = generating function for a Poisson distribution
Qb (s) = generating function for the screen's probability
distribution.-
Making the appropriate substitutions, the result is
N M n = I [ 1 smp (m) QaRAB PP ^ (28) n=o m=o where P^(m) = screen probability function
P (n) = Poisson distribution P Upon differentiating, one arrives at
3Qa b N _ M n-1 M in-1 = I n I smP c(m) I ms Ps (m) Pp (n) (29) 9s 5 n=o m=o iti= o 37
Evaluating at s = 1
N N n-1 M 3Q-AB s=i = I nl I- PSW I mPs (m) _ Pp(n) v (29) 9s n=o in=o m=o M But ^ P (m)=l, and thus m=o s
9Q N M AB = I I nPp (n) I I I mPs (m) \ (30) 9s s=l n=o m=o
Since
^AB, N = I nP(n) = n , 9s s=l n=o
the final result for the average is
9QAB (31) nAB nAnB 9s s=l
with a little manipulation, the expression for the variance of
the above system is:
IT TO IT! f I ? O ! O ’ab2 - % The average output n^g is also as expected, being the product of the average number of absorbed photons and the average gain per absorbed photon. Scintillation DQE With the above results, the can be given by (SNR)2 out nAn B (34) 2 Note that if the variance in the screen's gain, cR was zero, then DQEscint = 1, as one would expect. 39 Swank (1973) has shown that a screen's DQE ^ may be calculated if the moments of the screen's probability distribution were known. With this motivation, assume that P On) was obtained by observing the output for N absorbed x-ray photons. The moments can then be expressed as M = Yn = N o x n n M = Yn n 1 n n M2 = lNnn2 (35) n where = the number of absorbed x-ray photons that resulted in the output of n optical photons. If N is large enough, then DQEi may be represented by N 2 n DQE_. + = - ^ 5- ---- (36) scint N I - — 2 n Nn where use was made of the relation With a little algebra, the final result is ^scint = d Nnn)2 / N J N^n2 n n M 12 (38) M M o 2 Simulation of Screen Statistical Processes A rather interesting prospect lies in the possibility of obtaining in principle, an analytical output probability curve for a phosphor screen for a given input Poisson probability distribution and a given screen probability distribution. Upon inspection of Equations 22 and 28, it is evident that a problem arises when one tries to evaluate the quantity: The above involves a summation of a large number of terms, each term having a different coefficient, with the entire summation then taken to the nth power. For a measurement interval that corresponds to a Poisson average of 10, for instance, n may be as high as 15. However, some basic insights can be gained if one were to consider a screen distribution that is composed of only several dis tinct points. The summation will thus consist of only a few terms and the analysis could then be done easily with a computer. Two distributions were chosen for the simulation, as shown in Figure 20. The distribution in Figure 20 (a) is useful in that it somewhat approximates a screen distribution that exhibits k-escape. Dis tribution 20 (b) represents a scaled version of a CaWO^ distribution. The probabilities P^ and P^ were varied in the case of Figure 20 (a) and the Poisson input average of n^ was varied for both distributions. 41 C l . 0.2 0.2 5 10 3 6 9 12 No. of photons/event No. of photons/event Fig. 20. Simulation screen probability distributions. A Fortran program was utilized in carrying out the calculations. The results are presented in Figures 21-23. From the results, the following observations may be made: 1. The screen possesses a DQEscint > which degrades the signal to noise ratio as described in the previous section. 2. The DOE . decreases as the spread in the probabilities scint P. is increased. i 3. The DQEscint is independent of the Poisson average n^. 4. The output probability functions P ^ are rather as sym metric for small values of n^, but gradually approaches a symmetric form for larger n^'s. This may be explained by considering the shape of a Poisson distribution. 5. The probability of obtaining zero output in the measure ment interval is relatively high for small values on n^. Probability Probability 10 40 lln i. 2 Smlto rsls o Fg 2() P -/, ^ 3/4. - P^ -1/4, P^ 20(a), Fig. for results Simulation 22.Fig. i. 1 Smlto rsls o Fg 2() = 0.5. = 20(a), Fig. for results Simulation 21.Fig. AVG.-I7.50 v2 N - 1.373 SNR - ST.DEV.-12.75 - 20 80 1 — AVG.-I5.00 N - 1.342 SNR - ST.DEV.-II.1 1 -- Counts Counts 1 -- T~* 2 XI o Jl j 0 80 40 40 00 AVG.-52.50 V N - 2.370 SNR - ST.DEV.-22.07 AVG.-45.00 6 - n N - 2.324 SNR - ST.DEV.-19.36 A 6 1 ------i Counts 1 1 — ------Counts ------r *» JS X) o 0 80 40 0 80 40 AVG.-75.00 ° l - A n N =3. 0 SNR .0 3 = ST.DEV.-24.99 Count s AVn.-87.50 -10 n N - 3.045 SNR - ST.DEV.-28.74 A Counts Probability Probability 10 10 20 20 V. 6.10 AVG.- N - 0.8907 SNR - 6.85 ST.DEV. - Counts AVG.-24.40 N - 1.781 SNR - ST.DEV.-13.70 i. 3 Smlto rsls o Fg 20. Fig. for results Simulation 23.Fig. Counts o_ 10 10 20 20 AVG N - 1.260 SNR - TDV- 9.685 ST.DEV.- A 5 nA ■ Counts Counts 12.20 O- 10 10 20 20 AVG N - 1.543 SNR - TDV - 11.86 ST.DEV. - Counts Counts AVG.-36.60 N - 2.182 SNR - ST.DEV.-16.77 18.30 01 - p * Some Conclusions In summary, three important points should be reiterated. One, the signal to noise ratio of the output of an intensifying screen is not as predicted by Poisson statistics alone. Instead, it is degraded due to the variations in the scintillation output for a given absorbed x-ray photon. Secondly, if there are no variations in the scintillation output, the gain of an intensifying screen does not play a role in the output signal to noise ratio. However, gain is important in the consideration of the imaging devices that detect the screen's output, since noise and background from these devices may conceal the signal if the gain is too low. Thirdly, a screen's may be easily obtained if its probability distri bution is known. CHAPTER 3 EXPERIMENTAL PROCEDURE AND SETUP Introduction The basic experimental objective was to measure the photon emissions from x-ray intensifying screens and thus be able to obtain the statistical parameters of these emissions. As briefly touched upon in the introduction of Chapter 1, the experimental setup is a photon counting system consisting of a fast photomultiplier, wide bandwidth electronics, and a digital computer for data analysis and processing. An overall description of the system is shown in Figures 24 and 25 as is described by Roehrig et al., 1979. The phosphor screen is placed directly against the window of the photomultiplier in order to maximize the collection efficiency for the emitted light photons. The intensity of the x-ray source is low enough such that the overlapping of x-ray events is avoided. The absorption of a single x-ray photon results in the emission of several hundred to several thousand light photons, each light photon having a probability p of being absorbed by the photocathode. If a light photon is absorbed, a photoelectron is emitted from the photo cathode, which is then amplified by the subsequent dynode chain of 7 the photomultiplier. Amplification can be on the order of 10 , resulting in a measurable current pulse at the output of the 45 46 wwite m irz irxuAtmN «iaci - d D Fig. 24. Block Diagram of U of A evaluation facility. PHOTOELECTRON FAST SCOPE PUT PULSE 15 ns HEIGHT DISTRIBUTION ■^VX< 50 o 30 pF SCREEN PHOTOMULTIPLIER COLLIMATOR OYNODES DISCRIMINATOR NOISE COUNTABLE PHOTOCATHODE LEVEL PULSES PULSES QUANTUM E F F IC IE N C Y n TIME t 50 PHOTOMULTIPLIER PULSES PHOTOMULTIPLIER PULSES DUE TO THE ABSORPTION OF DUE TO THE ABSORPTION OF A A SINGLE X-RAY PHOTON FOLLOWING SINGLE X-RAY PHOTON 1 X-RAY PHOTON - 13 PMT PULSES I X-RAY PHOTON " 11 PMT PULSES Fig. 25. Schematic of PMT pulse counting method. 47 photo-tube'. Thus an absorbed x-ray photon will lead to a train of current pulses, each pulse to be amplified and counted by the associated electronics. The Photomultiplier The first and probably the most important stage of the system is the photomultiplier tube. It is an RCA 8850 with a bialkali photocathode and a GaP first dynode. The spectral response of the photocathode extends from 300 nm to 60 nm, and thus is coupled fairly well to the spectral outputs of the intensifying screens, particularly with the blue-emitting phosphors. The photomultiplier also has a high gain first dynode and an output pulse width in the order of several nanoseconds, both characteristics being necessary for photon counting measurements.. Details are given in RCA Photomultiplier Handbook (1970) con cerning the noise characteristics of photomultiplier tubes. One of the conclusions drawn from that discussion states that a photomultiplier provides noise free gain if the gain of the first stage is large, since most of the noise is from the first stage. The RCA 8850's GaP first dynode has a gain of 30 as compared to a gain of 3 for the remaining dy- nodes. Figure 26 is a plot of the pulse height spectrum of the photomul tiplier operated at 2140V. The illumination is provided with a green LED whose intensity is low enough for the resolution capabilities of the tube, i.e., the probability for multi-photoelectron events is negligi ble. As one can see, a single photoelectron peak is clearly 48 SINGLE ELECTRON 100 r ILLUMINATION: CREEN LEO NOISE DISCRIMINATOR SETTING OF 130 nV 200 LOVER LEVEL (nV) Fig. 26. RCA 8850 photomultiplier pulse height spectrum, resolvable. The discriminator level of the electronics is set at the valley separating the noise and the single photoelectron peak. The dark count of the photomultiplier is on the order of 160 counts per second, which is about an order of magnitude lower than the signal count rates that are obtained with the light outputs of the inten sifying screens. Thus, with a noise free gain and single photoelectron detection resolvability, the noise measured at the output of the detec tor is essentially that of the input. Figure 27 is a plot of the number of photoelectrons counted in a time interval of 100 msec with a 600 nm irradiance. The signal to noise ratio observed is that predicted by Poisson statistics, as expected. Another factor that must be considered is the count rate capability of the photomultiplier. In order to know what count rates Avg Mo of Counts per Measurement Interval = 2206 Stand. Dev. in the Mo of Counts = 47.6 2206 = 4? PMT Counts Fig. 2 7. Photomultiplier output corresponding to a Poisson distribution. are necessary, the decay characteristics of the light output from a screen should be examined. Assuming that the screen output decays as in Figure 2 8 and if the electronics to observe the decay is a simple RC circuit (Fig. 29), then the equation for solving the voltage output is dV -t/r + C = N e (40) o R dt Upon solving the differential equation, the voltage is given by 50 Screen Output No Decay Characteristics > " t/T . where t = time constant Time Fig. 28. Phosphor screen output decay characteristics TIK E j ( s ) 10 20 30 40 so 60 70 80 30 100 ICO 200 500 600 7oo 800 2 20 p f V(t) - - e"17') 7 • d e c a y tim e o f CaUO^ lu m in e sc e n c e Fig. 29. Decay time constant for CaWO^ screen. 51 For RC>>x, the rise time of the voltage signal is governed by the decay time t , whereas the fall time is determined by the RC time constant. Figure 29 also gives the plot of the voltage signal from a CaWO^ screen scintillation as a function of time, indicating a decay time constant of approximately 7 psec. The photon emission rate can be approximated with dt At t = 0, and supposing that Nq = 500 as suggested by Coltman, (1947) , a rate of ~ 70 MHz is obtained which would result in a 10 MHz count rate for detection efficiencies of 10-20%. The pulse width of a single photoelectron pulse at the output of the photomultiplier has been measured with the aid of a fast oscilloscope, giving a width of ~ 10 - 8 -1 nsec. The corresponding bandwidth is then given by Af = (2.10 sec) = 50 MHz. The counting rate capability of the tube is thus suitable for this measurement. The counting efficiency of the photomultiplier and its associated electronics was calibrated as shown in Figure 30. The UNIVERSITY or AUlZOVA FILTER ECO RADIOMETER PMT WWW FULSE COUNT CAMHA SCIENTIFIC STANDARD SOURCE Fig. 30. Calibration of photomultiplier counting efficiency. 52 source was a Gamma Scientific Standard Source RS10. With the aid of a set of 10 nm wide interference filters (Melles Griot No. 031FS005) the photocathode was illuminated with light of a known wavelength and radiant intensity. The intensity of the emitted x-ray flux was calibrated with a Hyperpure Germanium Detector (Ortec Model 1513) and its associated electronics (Ortec Amplifier 572 and Single Channel Analyzer 550). The measured intensities of the source are listed in Table 4 for the 3.2 mm colimator-aperture. Note that the intensities are low enough to prevent overlapping of the absorbed x-ray pulses and that even lower intensities are possible with smaller apertures. Finally, a typical radiographic x-ray generator Bremstrahlung source was also available. A Bremstrahlung spectrum is produced in this case and it is capable of being modified through varying amounts of irradiance was measured with an EG§G Radiometer-Photometer Model 550. Calculating the photon incidence rate at the photocathode and counting the number of pulses at the output of the photomultiplier- elect ronics system, the counting efficiency was obtained. This procedure was carried out for wavelengths of 500, 550, and 600 nm, with the efficiency for other wavelengths in the 350-600 nm range obtained through extrapolation. The extrapolation was done with the aid of a typical quantum efficiency curve for the RCA 8850. The result is shown in Figure 31. The effective counting efficiencies for the various emission spectra of the intensifying screens are Table 4. X-ray source characteristics. Target Mo Ag Du Ce Tb Tm ka (kcV) 17.4 22 32 34.5 44 50 Nko (counts/sec) 16.9 27.6 39.1 38.1 64.5 70 kg (keV) 20 25 36.8 39.7 50 58 Nkg (counts/sec) 3.2 5.5 10.7 10.5 18.4 18.4 .19 .20 .27 .28 .29 .26 V Nka ' COUNTING tFFICIiUNCY c.x 'c.cff w 0.05 >- * 5 s 0.0J wo s 0.02 I £ v 0.01 B 5500 60CO Fig. 31. Counting efficiency of the system. 54 obtained by evaluating [ E XT)cXdX ^c,eff = ------(43) E AdX where = screen emission spectra in energy density per unit wavelength ncA - counting efficiency of the system as a function of wave length. Table 5 gives the values of nc e££ for the various screens that were tested. The emission spectra in Figures 11-13 were utilized for these calculations. 55 The X-Ray Source The x-ray source is an Amersham Variable Energy X-Ray Source as illustrated in Figure 32. It contains a radioactive source, Am 241, whose gamma emissions strike secondary targets and cause characteristic x-ray fluorescence. Table 4 lists the six secondary targets and their respective characteristic x-ray spectra. Note that the emitted characteristic x-rays for a given target is not monochro matic, containing both k and k energies. filtration. Rotary taryet holder (5 targets) X-ray aperture O' 23 X.208 Dimensions in m m Fig. 3 2. Variable energy x-ray source. Electronics The associated electronics consist of an Ortec Amplifier 9302 with a gain of 20, an EG§G fast Discriminator TDlOl/N, and EG§G Logic Interface L 1380/NL, and a threshold counter that was built inhouse. Figure 3 3 illustrates the basic order for these components. After a current pulse from the photomultiplier is amplified and con verted into a voltage pulse by the amplifier, the discriminator deter mines whether this pulse is a signal pulse (Fig. 26). If the pulse is a signal pulse, it is then sent through the interface and into the threshold counter, where it is recorded and sent to the computer for analysis. The threshold counter is triggered by the rapid arrival of a series of signal pulses and counts these pulses within a given gate time. This gate time interval is selected such that it is greater than the total decay time of the phosphor screen for a single absorbed x-ray photon. The threshold counter may be also triggered independently by a pulse generator. This feature is useful in doing samplings that are to be independent of the signal. The bandwidth of the electronics must also be large in order to resolve the photoelectron pulses from the photomultiplier. The pulse-pair resolution for the Ortec 20X Amplifier is on the order of 9 nsec. The output pulse width from the differential discriminator is set to - 10.8 nsec. The bandwidth of the threshold counter is on the order of 20 MHz. Taking these components as a whole, the overall bandwidth of the electronics would be on the order of a little less than 20 MHz, sufficient for the 10 MHz requirement. Or tec EG&G EGGG AmplIfier DIfferentI a I Logic Interfact Threshold PMT 9302 Discr imi nator L1330/NL Counter 20X TDI01/N Signal and Noise Amplified Amp Iified Reshaped Current Pulses Signal and Noise Signal Pulses Voltage Pulses Voltage Pulses Fig. 33. System electronics. Computer Analysis After the threshold counter counts all of the pulses resulting from an absorbed x-ray photon, the computer records this as an x-ray event and assigns to it the number of pulses that were counted. This is carried out for each observed x-ray event, the data being stored in a one-dimensional array, HIST(K). The variable K is the number of pulses counted in an individual event and HIST(K) is the number of times an x-ray event resulted in K pulses. HIST(K) is therefore essentially the screen probability function P , as described in Chapter 2. The computer is also capable of recording the dark pulses from the photo multiplier and subtracting it from the above distributions. This ensures that all of the pulses that are counted and used in the analysis did indeed originate from an x-ray absorption event. CHAPTER 4 RESULTS AND CONCLUSIONS Results The probability distributions P^(m) for a single absorbed x-ray photon were obtained for a set of phosphor screens with varying incident x-ray energies and are shown in Figures 34-38. In the corner of each curve, the parameters of the distribution are displayed. They are the zeroth, first and second moments, the number of utilized x-ray events, the average number of counts per x-ray event, the standard deviation in the number of counts per event, and the calculated scintillation DQE. Also listed is the incident x-ray energy for each case (ignoring the k lines for the time being). For the case of the p Csl input phosphor of an image intensifier, the incident x-ray flux was produced with the x-ray generator with 2.6 and 8 mm of copper filtration. These particular spectrums of the input x-ray flux are shown in Figure 3 9. Figures 40-44 are plots of some of the above mentioned parameters as a function of incident x-ray energy,with the k-edge of some of the phosphors also being noted. Figure 4 2 is a plot for a series of ZnCdS screens with varying thicknesses. Table 5 lists the conversion properties of each of the screen as obtained with the above data. The counting efficiency of the 59 MO - 8129.00 MO - 12653.00 MO » 10053.00 Ml - 1266431.50 Ml • 556379.50 M2 • 45357976.00 Ml • 66)025.50 M2 • 156612672.00 MO. OF UTILIZED EVENTS M2 - 85541736.00 NO. OF UTILIZED EVENTS NO. OF UTILIZED EVENTS - 8129 ■ 12653 AVC. NO. OF COUNTS/EVENT - 10053 1 AVC. HO. OF COUNTS/EVENT - 68.47 AVC. NO. OF COUNTS/EVENT - 100.1 .O STAND. DLV. IN THE NO. OF - 05.85 •H STAND. DEV. IN THE NO. OF STAND. DEV. IN THE NO. OK COUNTS - 68.58 MO ■ 11418.00 mo • 10070.00 Ml - 1194394.00 mi - 1322273 00 MO " 9856.00 M2 - 151907630.00 M2 - 195506544.00 Mi - 1305801.00 NO. OF UTILIZED EVENTS NO. OF UTILIZED EVENTS M2 - 219445468.00 - 11438 ■ 10870 NO. OF UTILIZED EVENTS AVC. NO. OP COUNTS/EVENT AVO. NO. OF CCUNTS/tVENT •H - 9856 rH - 104.4 - 121.6 AVC. NO. OF COUNTS/EVENT •r-1 STAND. DEV. IN THE NO. OF STAND. DEV. IN THE NO. OF ■ 132.5 COUNTS - 48.75 COUNTS - 56.53 STAND. Dr.V. Ill THE NO. OF •s THE SCINTILLATION DQE. THE SCINTILLATION DQE. COUNTS - 68.64 & 0.8210 - 0.8224 THE SCINTILLATION D3E. o 0.7884 & 34 k.V 44 keV 50 keV 0) > •H P cd r H a> a; ~T~ I------I I 50 100 PMT Counts 50 100 PMT Counts 100 200 PMT Counts Fig. 34. Measured probability distributions (P ) for a o ZnCdS screen; Cd K-edge: 26.7 keV. Relative Probability , Relative Probability it 50 COUNTS - COUNTS M• 2• » < i 6III ' ) 9 M ■ 6050,00 1 D'iO.91 9 - 5 KO A VC. NO. OF COUflfJ/tVtNT OF A NO. VC. TN. t. NTE O OF NO. THE IN DtV. STAND. NO. OK UTI LI7tO SiWtlirjUTI OK LI7tO NO. THE SCINTILLATION 0Q2. SCINTILLATION THE l 3*165*49.13 • Ml 8*460.eo ■ MO 112 HO. OK UTILIZED EVENTS UTILIZED OK HO. TN. t. NTE O OF NO. THE IN DtV. STAND. COUNTS/EVENT OF NO. AVC. ONS- 17.79 - COUNTS H SITLAIN DQE. SCINTILLATION THE • 0.7388• - 6050 - • • 0.8684 - *45.69- ■ 23.36 8*460 2033 PMT Counts PMT PMT Counts PMT 17 keV 13.89 3*4 kV * 6 6.00 169 50 . Fig. 35. Measured probability distributions (Pg) for a for (Pg) distributions probability Measured 35. Fig. aO sre; -de 6. keV. 69.5 - k-edge W screen; CaWO^ •H •H rH & r I •s ■P & cd > o 0 O) ) I 50 tMju ONS 13.92 * COUNTS 2y973.3l •Ml TN. t. NTE O OF NO. THE IN DtV. STAND. T N L V A S T H CU C OF A NO. VC. 0070913.00 - M2 Tilt SCINTILLATION DQE. Tilt SCINTILLATION EVENTS UTILIZtU OK NO. 70*111.00 • MO 0.02*49- 30.21 ■ 70'i*»■ ONS» 19.66 » COUNTS 2*11 - 2 M 20o*i6 332503.00 • Ml . 00 THE SCINTILLATION DQE. SCINTILLATION THE OF NO. THE IN DEV. STAND. 6026.00 • MO AVC. NO. CF COUNTS/EVENT CF NO. AVC. NO. OF UTILIZED EVENTS UTILIZED OF NO. 0.8084 - =» 56.0*4 ■ 6026 PMT Counts PMT 2 k«V 22 PMT Counts PMT 44 k#V •H rH K & l r • •H rH i? > 0 01 50 2 19951300.00 - M2 *416761 - Ml .22 ONS- 15.90 - COUNTS AVC. NO. OK COUNTS/EVENT OK NO. AVC. TN. E. NTE O OF NO. THE IN DEV. STAND. NO. OK UTILIZED EVENTS UTILIZED OK NO. 9961.00 ■ FO niE SCINTILLATION DQE. SCINTILLATION niE COUNTS = COUNTS THE SCINTILLATION DQE. SCINTILLATION THE TN. t. NTE O OF NO. THE IN DtV. STAND. AVC. NO. CK COUNTS/EVENT CK NO. AVC. 2= 30931 39*4.00= M2 433095.41 - Ml 7141.00 - MO NO. OK UTILIZED EVENTS UTILIZED OK NO. 0.8738 ■ 9963 ■ *41.83- • • 61.35 - «■ 7141 PMT Counts PMT PMT Counts PMT 0.6689 2 v k 32 23.83 50 keV 50 Relative Probability , Relative Probability I 1 1 50 100 PMT Counts 100 PMT 50 50 100 PMT Counts 100 PMT 50 r ------O• 4k V k 34 cen B keg: 75 e. 4 5 keV. 37.5 k-edge: Ba screen; . t. a ... 1 50 100 PMT Counts 100 PMT 50 50 100 PMT Counts PMT 100 50 1 r " I l 546000.00 - Ml 2• 59065140.00 • M2 O- 6473.00 - MO ONS 44.61 " COUNTS NO. OK UTILIZED EVENTS UTILIZED OK NO. TN. E. NTE O OF NO. THE IN DEV. STAND. AVC. NO. OF CCUNTS/EVENT OF NO. AVC. H SITLAIN DQE. SCINTILLATION THE :i , l?! . . . - 0.8411 - | l 153437.59 - Ml 4'3'»).00 • MO 2 JO'/'iV/lli.OO • M2 ONS 31.48 = COUNTS AVC. NO. OK COUNTS/EVENT OK NO. AVC. EVENTS UTILIZED or NO. TN. u. N H N. Of NO. THE IN UuV. STAND. THE SCINTILLATION DQE. SCINTILLATION THE - 84.47 • ■ ■ 6473 0.7820 - 72.42 - 4963 4 keV 44 2 k*V 22 L luii,i i? i i W i i L i i t i ! 50 I O- 7466.00 - MO ONS- 44.42 - COUNTS 2- 79596976.00 - M2 695070.44 • Ml H SITLAIN DQE. SCINTILLATION THE TN. DI.V. OF Ill NO. STAND. THE COUNTS/EVENT OF NO. AVC. NO. Or UTILIZED EVENTS UTILIZED Or NO. 100 PMT Counts PMT 100 100 PMT Counts PMT 100 I ' H | COUNTS . COUNTS THE SCINTILLATION DQE. SCINTILLATION THE TN. c. N H N. OF NO. THE IN DcV. STAND. AVC. NO. OF COUNTS/EVENT OF NO. AVC. 2• 3f»552024.OO • M2 l 336573.59 - Ml NO. OK UTILIZED EVENTS UTILIZED OK NO. 3740.00 - MO 0.8148 * 93.21 - 7466 » 89.99 • 3740 ■ 0.8766 . 50 keV 50 33.76 32 keV 32 to O' i i Relative Probability 0 0 ; SO L;',; SO L_...Ly " XTItUIIO* DC S ?! « !« .? • IS DQC * O I I U t I T lX K x t * Itll.I.N oo N . I . l l t I i* H ) » ai . is ia v t I * a - I s i u i l l U o» V) WC.. Fig. 37. Measured probability distributions (Pg) for a for (Pg) distributions probability Measured 37.Fig. 100 100 J OUISAVie OAI VO UIIIS/AIVViMe LO MJ M ‘1 f OMS1X61 I Al • SCINTIllAtlON ll rn A 4 I lit* uqi IS -OUMTS/11X16414 Of ‘01 AM. M l. 1 :6 1 4 HI HI 0,1 .. 6 4 .4 1 0 4 :6 1 4 1 l. M l. M nv\». nv\». O O UTHCI Al tt. S • |>e • tS .x t.t ll A ( ID C H T U Of SO. S POTCowt, t w o C T O P ISO ««ioie 150 dii PJfT Count* 04 in n, . of cot its no, • «. dkeg: 02 keV. 50.2 k-edge: Gd aCSG- sre; Lak-edge: screen; S La^CLS-Gd-O 34.7 lev 0 /" I. i - SO so 100 . V? 114 ? lV M 11644.00 I. H 'II. 64.. 00 .4 . 4 :6 4 6 . . 0 o :iui:Mi i M : i u i : oo i « tie 'X ik iii.u tiu N i«;i is is i«;i N tiu iii.u ik 'X tie suuv, s: t i N O01 0 ( 1 1 1 : 1 4 vi m of ni»66u*io iio * u 6 6 i/» in u o c f o SIAMl m Avti. T I E S f I M T I L L A T I O N I f H I S4 ' iso f I I H l l k / A C . ' W t l in DLV. tie vi i * - i a 2 kcV .2 2 2 1:1 Ai s* is iM lA «AAI S kcV .S 4 4 TIE ur MW 4 tmihT* ' !• all*. I HAT < :111 (VINTS. 38.9 Of rowns 1 6 mi m *4 keV, 5 0 so 100 T l * J C I . V T I U A T I 0 4 t A } t I« S TIE SCII.TILLATION SCII.TILLATION TIE oo : 04 * 14011,, :.|is:4i M l. M l. . f Nr ABVIfO % AI . I fA % IfflO V B /A rs lN U o lf I I. V ll.. A siAvi oiv »> vi of imii:in s.mi ikivr,. s.mi ofimii:in vi .111160 .»> HI. 11 4 *4 14. . t4:i:„o .w t4:i:„o . ISO 46,14 PSTT Counts in 00 no vo of 001 S 1:14 4 IS toons . |;|l 0-4 04 N O Relative Probability k-Cscape tk I Ttak Mil 10 20 i. 8 Maue rbblt itiuin fraCl x-ray Csla for distributions probability Measured 38.Fig. 3l>>rr Counts f i n TN C IN TH£ Of NO.COUNTS STAND - DC* Of SO.COUNTVAIStieitO *»C. • T l U SII90So 7S HI. O OF I7TILI:C0HO. 6VCVTS «-M* • hi O Sill DO HO- . n o i t u i i t n i c s I ■5 I I .*4 i t o o : » i l l l l l l l l . l l 59 keV.35.9 34.7 kcV 34.7 mg itniir I -de 3. e, s k-edge: Cs keV, 33.2 I k-edge: intensifier; image y i io< i;o o is oyc the STAND III* IN (IU NO. I# COUNTS I# NO. (IU IN III* STAND «« sti.HitiATiow oqe IS of uTin:te i-a*fof uTin:te tvfvri. 3fi*irr Count t "5 5 0 oo 22.2 koV 22.2 a.fits 23 THCi:MTIl.lATI0X 0 IS.2 DQC IS *. O Of COUNTS/AtsoaICO NO A*C. I-SAT TA D D* IN TieST DC*AND. ^-MTCounts Of I UTILCCD SAT STINTS *AS07.'0 oo Jll 10 44.S kcV 44.S . M Of COUNTS 20 3f, l 11700$ Ml. C SSSfill1C* IS NO V. O Of COUNTS/AFSONHJ S-OA* NO IVC. T ie SCINTILLATION SCINTILLATION ie T IN Till DC*. 01 COUNTSSTANDNO. I PSfT Counts 00 IIS* Of 21.11 I V M U U O I - U T I I7 00SIS7 Illillll.U 001 32.2 kcV 32.2 S a SO.* a IS TINTS si. O Of IUTILCCD SATNO CVCXTS- :I9»S0 74*1.00II Ml. HO- TieSCINTILLATION 01*27 DQCIS ; 7413610M;. DO $ A OfVC. COUNTS/ASSOf NO ICO: SAT I t 3S'HT Counts vo. on on vo. 00 11.71 :* ;* : n me me no or keV e k 7 . O S counts • ■t.O' i . 9 Besrhug pcrm fr ifrn aons of amounts different for spectrums Bremstrahlung 39.Fig. Re 1 a t i vc In tens i Ly filtration. 20 .m Cu 2.6mm 0 kVp50 40 U-K 60 -a Eeg (keV) Energy X-Ray 80 0 kVp80 100 65 66 No. of Counts per Event TmQ 60 40 20 Mo 0 Absorbed X-Ray Energy (keV) Fig. 40. Measured output emissions for a CaWO screen. 67 No. of Counts per Event 150 Tm Tb Bao 100 Tm k-Escape Tb k-Escape 0 20 40 60 Absorbed X-Ray Energy (keV) Fig. 41. Measured output emissions for a BaSO. screen. 68 No. of Counts per Event 200 150 100 Cd k-edge 10 30 50 Absorbed X-Ray Energy (keV) Fig. 42. Measured output emissions for a series of ZnCdS screens. Mean Count Averaged over Full Di sir i but ion 100 0 Counts at Full tncrgy Peak / a Counts at / o k-oscape peak / 80 / / 60 >. m k-Escapc oc / DQE i scint x /o tn 4-f La k-edgc z Scinti Hat ion Hat DQE ion Scinti 20 30 A0 50 60 X-Ray Energy (XcV) Fig. 43. Measured output parameters for a La^O^S-Gd^O^S screen. o '-O PMT Counts/Event 20 30 >\0 10 i. 4 Maue upt aaees o s xryiae intensifier. image x-ray Csla for parameters output Measured 44.Fig. hmsn CSFThompson Intens i f i e r 0 10 F u l l DiF u l l s i r i b u t i p n - ^ en Counts/EventMean Averaged overAveraged 20 tpo" i- Cut/vn from □Counts/Event oFull from Counts/Event Eg I "Edge a Eeg (KeV) EnergyKay -sae Peakk-Hscape nry PeakEnergy Scinti11 at ion DQE Cs-Edgc /|0 J. , -fo- .7 .8 .9 Scintillation DQE o Table 5. Measured output efficiencies of some phosphor screens. Phosphor No. of Counts No of photons A Energy Emitted Luminescent £c,eff absorbed keV "o emitted eff Energy Absorbed Radiant absorbed keV (nm) % Efficiency*** ZnCdS 3.25 4.7 69.15 535 16.0 18.0 CaW04 1.14 11.6 9.83 435 2.80 3,5,5.4** BaS04 2.88 15.7 18.34 388 5.86 6 (La202S Gd202S) 2.10 2.3 91.3 572 19.8 15* * For a GdO^ screen ** Intrinsic Conversion Efficiency (Coltman, 1947) *** Stevels, 1975 72 system is taken into account and the results are compared with those that are available in the current literature. Finally, Figures 45-46 are plots of the signal to noise ratio as a function of the average number of absorbed x-ray photons in the measurement interval. The data was taken with the Csl image intensifier and with a CaWO^ screen. The x-ray input for the Csl phosphor was as described above, whereas the input for the CaWO^ screen was obtained with a high intensity 44 keV source that is similar in nature to the variable energy source. Analysis Upon inspection of the distributions, it is rather obvious that the nunber of UV and light photons emitted is not constant for each absorbed x-ray photon. There are also many cases where there are more than one peak in the distribution, the cause being the k-fluores cence of the phosphor material. Note that a double-peak distribution always occurs when the energy of the incident x-ray photon is above the k-edge of the phosphor material. Of particular interest are the distributions obtained with the BaSO^ screen. The principal absorber in this case is Ba, which locates the k-edge at 37 keV. Note that when the secondary target of the source is Ce, there is still a second peak although the k^ line for Ce is at 34 keV. Obviously the k-fluor escence is caused by the k^ line of Ce, which is at 38 keV and is therefore above the k-edge of Ba. From the relative heights of the two peaks for the various distributions, it appears that a significant portion of the k x-rayS Fig. 45. Measured signal to noise ratios from the output of a Csla of output the from ratios noise to signal Measured 45.Fig. Signal to Noise Ratio 10 a tigrdb inl pulses. signal by triggered was the threshold counter did not sample continuously but continuously sample not did counter threshold the bobdxrypoos nerrdet h fc that fact the to due inerror photons 10x-ray absorbed below values experimental intensifier image x-ray siae N. fAsre XRy Photons X-Ray Absorbed of No. Estimated 10 Obtained from Distribut from Obtained xeietl S/N Experimental 100 Ideal S/N, assuming assuming S/N,Ideal oso Distribution Poisson j i _ j j . j _ i ons IzJ - j 74 SNR 4 3 Poi sson 2 Measu red SNR 2 6 8 10 No. of Events/Interva 1 Fig 46. Measured signal to noise ratios from the output of a CaWO screen; 44 keV incident x-ray photons. Threshold counter sampled continuously, triggered by pulse generator. 75 do escape from the screen (as predicted by Vyborny in Chapter 1) and thus reduces the efficiency of the screen. The positions of the peaks have been calibrated, indicating that the peak to the right corresponds to the case of the x-ray being reabsorbed and thus represents the full utilization of the absorbed energy. The peak to the left is a consequence of a k-escape and therefore represents only partial utilization. The calibrations are plotted in Figures 40-44. The results and comparisons presented in Table 5 are somewhat encouraging. First of all, one must realize the difference between intrinsic conversion efficiency and output efficiency, the latter taking into account optical losses within the screen whereas the former does not. The data presented for the CaWO^ screen appears to agree quite well with that proposed by Coltman (1947). The measured light output indicates an output efficiency of 2.8%, which in turn represents an intrinsic conversion efficiency of ~ 5.6% assuming optical losses of 2 ~ 50% as suggested by Coltman for a screen of thickness 84 mg/cm . The data for the other screens also agree quite well with the data available in current literature. The only notable discrepancy appears to be with the La^OgS-GdgO^S screen. It is noted however, that the counting efficiency of the system is quite low for this particular screen, being at the tail end of the system’s counting efficiency curve (Fig. 30). Errors in the calibration measurements can therefore easily account for the above discrepancy. Future investigations are obviously needed in order to remedy this situation. 76 One of the more important pieces of information obtainable from this data is the DOE . . values for these various screens. scant Two conclusions can be drawn. First, the effect of having the phosphor k-edge energy is to lower the scintillation DQE, as one would expect. Apparently, the attempt to enhance the absorption capabilities of intensifying screens has resulted in a slight reduction of the scin tillation DQE. Secondly, one can conclude that the observed ^Q^s c ^n .(-s are all rather high, the lowest being 73% and the highest 90%. This would indicate that the degradation of the signal to noise ratio after absorption and amplification is relatively small. In an attempt to observe the effects of the scintillation DQE, the data as shown in Figures 45-48 was obtained. The phosphor screen was irradiated with a high incident x-ray flux and the light output of the screen was sampled continously with a certain sampling time and area. If the screen had a DQE of 100%, then the set of sampled data should correspond to that predicted by Poisson statistics. The average number of absorbed x-ray photons in the sampling time is varied through altering the rate of absorption and by fixing the dura tion of the sampling time. For the data measured with the Csl image intensifier (Figs. 45, 47) the results are as predicted when the number of absorbed x-ray photons in the sampling interval is high. However, the results are not as good when the number of absorbed x-ray photons is low. This can be accounted for by considering the trig gering of the threshold counter as discussed in the section of Chapter 3 concerning the electronics. At the lower rates of absorption, the ML). 4-110.00 Ml. 4014:4 SO (42 * 44.010:0.00 >0 Of UlllUEO 1-IAT (VESTS* M l 4VC° NO Of COUNTS/A4S04ILD l-IAT • ST*.**0 OtV. IN TIH » - O f COUNTS * IN TIC NO. O f COUNTS • 4 1 . lt 2 X-«;iy 4 X -R a y PhutOHS l Photons PMT Counts 200 4 0 0 0 PMT Counts Mil- 1191. iNI /■>• Uvi'> VU m i* i: 4 4 : v i so M l- #744.14, W) N:*s:'(.s4-i„ ,N( Ml* ituu i to .^OO 0i 0T' l:;tD '-VT MINTS. * m :*;j7i,*:s'*.oo 11 :.0 01 U T IL K IU K #AV SViHTSe NO. O f UTILCIO 1 -IA T tVIVTS* or coaNTs/iisaLiri, v m i . 4 00 ivc;. NO or COUNTS/IkMlMtrO X -M f • *1 6 NO Of COUNTS/USONlfO 1 -IA T • ST.VIH DfV IN n it VO C f COUNTS . 11*.f STAND DCV. IN T ilt NO Of- CJJKTS • STAND77 ** 01V. IN TIC NO Of COUNTS • 10 X -R a y 8 X -R a y P h o to n s 6 X-Ray P h o to n s Photons til m iitii .Ai-Liu-i 200 4 0 0 4 0 0 0 200 PMT Counts 400 PUT Counts PMT C o u n ts Fig. 47. Csl image intensifier output distributions. HO • 3110.00 HO • 2956.00 h i HO • 3079.00 Ml - 316666.00 Ml • 453011.00 1 HI • 140000.00 f i H2 • 47S647T6.00 M2 • 91510123.00 H2 e 12412084.00 .r4 NO. OF UTILIZED EVENTS NO. OF UTILIZED EVENTS NO. OF UTILKID EVtSTS H • 1110 • 2956 • 3079 •'"I AVC. NO. OF COUNTS/EVENT AVC. NO. OF COUNTS/EVENT Id AVC. NO. OF COUyTS/EVENT X] - 94.27 • 145.19 & - 42.47 MO ■ 2630.00 MO - 2280.00 MO • 2400.00 HI • $43481.7$ Ml • 584619.19 Ml • 766744.00 M2 « 119114640.00 $ M2 ■ 180251160.00 M2 • 286421872.00 NO. OF UTILIZED EVENTS •H NO. OF UTILIZED EVENTS I—I NO. OP UTILIZED EVENTS • 2680 •H • 2230 • 2400 AVC. NO. OF COUNTS/EVENT AVC. NO. OP COUNTS/EVENT AVC. NO. OF COUNTS/EVENT • 191.11 > 240.64 • 102.14 n) STAND. DEV. IN THE NO. OF s t a So . d e v . in the n o . or .0 STAND. DEV. IN THE NO. OP COUNTS • 101.71 o COUNTS • I IS. 26 COUNTS • 111.25 k 0) > • f-l 4J cti r~t a> n: LiLilLU PMT Counts <100 ri.1T Counts HIT Counts Fig. 48. CaWO^ screen output distributions. -~i 00 79 counter was not triggering continuously and independently of the signal, but was triggered whenever an x-ray photon was absorbed and detected - thus the signal to noise ratios obtained would not be governed by Poisson statistics at all. This problem was remedied by trigger ing the counter independently with a pulse generator. A CaWO^ screen was utilized and the results are shown in Figures 46 and 48. The signal to noise ratio is indeed lowered as predicted by the scintillation DQE. Output distribution plots (P^g) relating to both sets of data are displayed in Figures 21-23. There appears to be a very good correlation between these experimental distributions and those that were created with the computer simulations as de scribed in Chapter 2. Conclusions On the basis of the theoretical and experimental analysis thus presented, three basic conclusions may be arrived at: 1. The output efficiencies of the newer rare earth screens are indeed superior to that of the more traditional CaWO^ screens. 2. The observed DQE . . values of the screens are high, scint indicating only a small degradation in the signal,to noise ratio. . 3. The decrease in the DQE . . of most of the newer screens scint due mainly to k fluorescence is rather small, differing from the DQE values for CaWO^ by only several percentage points. As a final remark, a very important parameter in the analysis of image formation has been neglected in the discussion thus far. There is still the need to evaluate the modulation transfer functions (MTF) of the newer screens. It can be an interesting and important factor to consider since the k-fluorescence which decreased the DQEsG^nt may also decrease the resolution capabilities. If the degradation in resolution is found to be small, then it may be unimportant since the readiologist is the final component in the radiographic process and very fine spatial ,detail may not be that important for the diagnosis (Rossmann, 1974). REFERENCES Alves, R. V., and R. A. Buchanan, "Properties of Y-CLSrTb X-Ray Intensi fying Screens," IEEE Transactions in Nuclear Sciences, Ns-20 415 (1973). Bates, C.-W. Jr., "Scintillation Processes in Thin Films of Csl (Na) and Csl (Tl) due to Low Energy X-Rays, Electrons, and Photons," Advances in Electronics and Electron Physics, 28A 451 (1969). Beckman, S., L. H. J. F., A. J. Vermeulen, "The Electrodelca, A New Photofluorographic Camera with Image Intensification," Paper presented at the Fourth European Electro-Optics Conference (EEO 78), Utrecht, Netherlands, (1978). Buchanan, R. A., S. I. Finkelstein, and K. A. Wickersheim, "X-Ray Exposure Reduction Using Rare Earth Oxysulfide Intensifying Screens," Radiology 105 185 (1972). Cleare, H. M . , H. R. Splettstosser and H. E. Seemann, "An Experimental Study of the Mottle Produced by X-Ray Intensifying Screens," American Journal Roentgenology, Radium Therapy Nuclear Medicine . 88 168 (1962). Coltman, J. W., E. G. Ebbighausen, and W. Altar, "Physical Properties of Calcium Tungstate X-Ray Screens," Journal of Applied Physics 18 530 (1947) . Kingsley, J. D., "X-Ray Phosphors and Screens," General Electric Corporate Research and Development, Schenectady, N. Y. (1975). Ludwig, G. W. and J. S. Prener, "Evaluation of Gd202S:Tb as a Phosphor for the Input Screen of X-Ray Image Intensifiers," IEEE Transac tions in Nuclear Science, NS-19 3 (1972). Messier, R. F. and R. W. Wolfe, "Engineering Aspects of Recent Phosphor Developments in X-Ray Intensifying Screens," presented at Penn State University, (1973). | ' Nudelman, S., M. M. Frost, and H. Roehrig, "Image Tubes and Detective Quantum Efficiency,V Advances in Electronics and Electron Physics Academic Press, N. Y. 4OB 539-551 (1976). 81 82 Rao, G. , P. Fatouros and A. E. James, "Physical Characteristics of Modern Radiographic Screen-Film SystemsInvestigative Radiology JL3 460 (1978) . Rao, G., and P. Fatouros, "Quantum Fluctuations in Radiographic Screen- Film Systems," Medical Physics 6 118 (1979). RCA Photomultiplier Handbook, RCA, Commercial Engineering, Harrison, N.J., 07029, (1970). Roehrig, H . , B. Lum, and C. Dick, "Measurement of the X-Ray Induced Light Photons Emitted from Radiographic CaWO Intensifying Screens" Proceedings SPIE, Application of Optical Instrumentation in Medicine VII, Toronto, Canada (1979). Rossi, R. P., W. R. Hendee, and C. R. Ahrens, "An Evaluation of Rare Earth Screen-Film Combinations," Radiology 121 465 (1976). Rossmann, K., "Spatial Fluctuations of X-Ray Quanta and the Recordings of Radiographic Mottle," American Journal of Roentgenology, Rad. Therapy, and Nuclear Medicine 90 863 (1962). Rossmann, R., "An Approach to Image Quality Evaluation Using Observer Performance Studies," Diagnostic Radiology 113 541 (1974). Shaw, R., Editor, "Selected Readings in Image Evaluation," Society of Photographic Scientist and Engineers (1976). Society of Photographic Sci. § Eng., SPSE, Handbook of Photographic Science and Engineering, John Wiley G Sons, N. Y. (1973). Stevels, A., "New Phosphors for X-Ray Screens," Medica Mundi 20_ 12(1975). Swant, R. K ., "Absorption and Noise in X-Ray Phosphors," Journal of Appl. Phys. 44 4199 (1973). Vybomy, C. J . , C. E. Metz, K. Doi, and A. G. Hans, "Calculated Charac teristic X-Ray Reabsorption in Radiographic Screens," Journal Appl. Photographic Engineering £ 172 (1978) . Wagner, R. F., "Toward a Unified View of Radiological Imaging Systems. Part II:Noisy Images," Medical Physics 4 (1977). Weidner, R. T., and R. L. Sells, Elementary M o d e m Physics, Allyn and Bacon, Inc., Rockleigh, N. J ., (1973). 3S 1