MEASUREMENT OF DIAGNOSTIC X-RAY SPECTRA USING CdZnTe DETECTOR

Transcription

1 MEASUREMENT OF DIAGNOSTIC X-RAY SPECTRA USING CdZnTe DETECTOR M.Matsumoto 1,S.Miyajima 1,A.Yamamoto 1,T.Yamazaki 1 and Kanamori 2 1 Sch. of Allied Health Sciences, Osaka Univ., 1-7 Yamadaoka, Suita, Osaka Japan 2 Dep. of Electric Engineering,Fukui Univ. Tech.,3-6-1 Gakuen,Fukui Japan INTRODUCTION We directly measured diagnostic X-ray photon spectra produced by a single-phase 2-pulse diagnostic X-ray unit driven at 50 to 100 kv and a few ma(1), a mammographic X-ray unit at 25 to 32 kv and 3 times of 80 mas(2) and an X-ray computed tomography (CT) scanner at 120 kv and 3 ma(3) have been analyzed. Since detected spectra were distorted by the response of CdZnTe(CZT) detector and did not present the true photon spectra, the correction by the stripping procedurec(4) was applied. The response function of detector used in this procedure has been determined by an evaluation of interactions (K-escape, coherent scattering, Compton scattering) and incomplete charge collection calculated using the Monte Carlo method(4-7). Castro et al.(7) have attempted to perform the direct measurement of diagnostic X-ray spectra using a low-efficiency CdTe detector. They analyzed photon spectra produced by a portable diagnostic X-ray spectra unit from 45 to 100kV. They have used only an average value of K α (25 kev) radiation on account of the little difference between Cd(23 kev) and Te(27 kev) absorption coefficients and have not considered K β radiations of Cd and Te. They have used the constant value for the incomplete charge collection and have not considered Compton scattering. Therefore their correction of spectra have been bad accuracy. But we have used K α1,k α2,k β1,k β2 radiations of Cd,Zn and Te respectively and have used the weight function for the incomplete charge collection and have considered Compton scattering. The Monte Carlo simulations were continued with changing the important factors[mean path length of hole λ h, dead layer of CZT crystal (3 3 2 mm 3 ) and weight factor W q ] of incomplete charge collection until the fittest response function was found out. Therefore our correction of spectra have improved the accuracy. Corrected photon spectra were compared with the diagnostic X-ray spectral data published by Bureau of Radiological Health (BRH) measured using a Ge detector. Exposures were calculated from the corrected photon spectra and compared with exposures measured using an ionization chamber. METHOD 1. DIRECT MEASUREMENT OF DIAGNOSTIC X-RAY SPECTRA We have directly measured diagnostic X-ray spectra using the Ge (PRINCETONGAMMA-TECH NIGP ) and CZT (TOYO MEDIC RAMTEC 413) detectors. Photon spectra transmitted through an object(1 mmal to 6 mmal) produced by a single-phase 2-pulse diagnostic X-ray unit (SHIMADZU UD150L-5 and CIRCLEX 1/2U13CN-25) driven at 50 to 100 kv and a few ma have been analyzed(1). Each exposure attenuation curve for 50 to 100 kv is calculated and is compared with the curve measured by an ionization chamber (VICTOREEN RADCON MODEL 500 and TYPE 0.1MA PROBE). 2. DIRECT MEASUREMENT OF MAMMOGRAPHIC X-RAY SPECTRA We have directly measured mammographic X-ray spectra using a 35μm-collimator and the CZT detector. Photon spectra transmitted through 0.03 mm Mo or mm Rh filter and object (0.1 mm Al to 1.2 mm Al) produced by a mammographic X-ray unit (GE SENOGRAPHE DMR) at 25 to 32 kv and 3 times of 80 mas have been analyzed(2). Each exposure attenuation curve for 25 to 32 kv is calculated and is compared with the curve measured by an ionization chamber (PTW-FREIBURG TYPE 23344). 3. DIRECT MEASUREMENT COMPUTED TOMOGRAPHY(CT) X-RAY SPECTRA We have directly measured computed tomography (CT) X-ray spectra using the CZT detector. Photon spectra transmitted through object (1 mm Al to 8 mm Al) produced by a CT scanner (HITACHI LSCT) driven at 120 kv and 3 ma have been analyzed(3). The exposure attenuation curve for 120 kv and 3 ma is calculated and is compared with the curve measured by an ionization chamber (VICTOREEN RADCON MODEL 500 and TYPE 0.1MA PROBE). 4. CORRECTION OF MEASURED X-RAY SPECTRA 4.1 CALCULATION OF INCOMPLETE CHARGE COLLECTION The response functions of CZT detector are obtained by the simulation using Monte Carlo method(4-7) to correct measured spectra obtained with the detector. Monte Carlo method are a well-established method

2 for simulating X-ray photon interactions (K-escape, coherent scattering and Compton scattering processes) in the detector using the random number generator. For an incident X-ray photon, the depth of the initial interaction is determined from the total detector crystal cross section. One of the three processes is then selected on the basis of their relative cross sections by random-number sampling. The simulation is continued until the total energy of the X-ray photon will be absorbed as the charge carriers(pairs of electrons and holes) in the detector. If the charge carriers drifting in the detector are trapped and can not reach the collecting electrode, the output pulse height decreases. In this simulation the charge carrier trapping in the CZT detector had to be considered as follows. For a planar type detector, the collected charge Q(t) on the electrode is represented by Q(t) = (q 0 /d) [drift distance of electron + drift distance of hole], (a) where d is the thickness of the detector, and q 0 is the charge carriers created by photon interaction. In Fig. 1, FIG. 1. Simulation model considering the carrier trapping in the CdZnTe semiconductor detector. are created at the interaction position and the distance from this position to the cathode is indicated by x. The the incident X-ray transmitted through the dead layer which is an inactive layer for X-rays, and the charge carriers average collected charge, which is dependent on the position of interaction x, is described as Q(x). The value of Q(x) has twoequations according to the length between the mean free path λ h of hole and x. The mean free path λ h of hole is an average length of holes passed without collisions to electrons. For x λ h, For x > λ h, Q(x) = q 0 Q(x)=q 0 [1- W q (x - λ h )/d ] (b) (c) Where W q is weight factor. The energy depositions in the CZT detector calculated by Monte Carlo method are multiplied by the weight function, which is defined in the square parentheses of equation (c). Therefore, except for the condition [1], the energy multiplied by the weight function is less than the energy when the charge carrier trapping is not considered. In this way, response functions can be simulated considering the charge carrier trapping in the CZT detector as stated in FITTEST MONOENERGETIC RESPONSE FUNCTION Monoenergetic response function and full energy peak efficiency for the CZT detector have been determined by an evaluation of interactions and incomplete charge collection using Monte Carlo method elaborated on a NEC personal computer PC-9821 Ra20. In this simulation, important factors are mean path length of hole λ h, dead layer of the CZT crystal and weight factor W q for incomplete charge collection. This simulation are continued with changing the important factors(λ h =0.005 cm to 0.15 cm at every cm interval, dead layer=0.01 cm to 0.10 cm at every 0.01 cm interval and W q =0.1 to 1.0 at every 0.1 interval) until the fittest response function is found out. The values of factors are decided from the following reasons; (1) the calculated full energy peak efficiency is coincided with measured full energy peak efficiency using isotopes( 241 Am, 109 Cd and 57 Co) as shown in Fig.2, (2) the corrected spectra for the CZT detector are coincide

3 FIG. 2. Comparison of calculated full energy peak efficiencies for any factors and measured full energy peak efficiencies using isotopes( 241 Am, 109 Cd and 57 Co). FIG. 3. Comparison of (a)measured and (b)corrected γ-ray spectra of an isotope 241 Am for Ge(solid line) and CZT(dotted line) detectors. FIG. 4. Comparison of exposure attenuation curves obtained from the CZT spectra and the ionization chamber at (a)70 kv and (b)100 kv. with those for a Ge detector as shown in Fig.3, and (3) exposure attenuation curves calculated from corrected spectra for the CZT detector are coincided with those measured using the ionization chamber as shown in Fig.4. For example the monoenergetic response functions for the CZT detector with λ h =0.025 cm, dead layer=0.06 cm and W q =0.3 obtained at 50, 100 and 150 kev are shown in Fig.5 respectively. The full energy peak efficiencies obtained for the Ge and CZT detectors are shown in Fig. 2. The efficiency for the CZT detector

4 rapidly decreases at higher energy ranges than 20 kev. Therefore the photon spectra in high intensity X-ray beams can be measured by the CZT detector. 4.3 SRRIPPING PROCEDURE Castro et al.(7) have used only an average value of K α (25 kev) radiation on account of the little difference between Cd(23 kev) and Te(27 kev) absorption coefficients and have not considered K β radiations of Cd and Te. They have used the constant value for the incomplete charge collection and have not considered Compton scattering. Therefore their correction of spectra has been bad accuracy. But we use FIG. 5. Monoenergetic response functions obtained at 50, 100 and 150 kev for the CZT detector with λ h = cm, dead layer = 0.06 cm and W q = 0.3. K α1,k α2,k β1,k β2 radiations of Cd,Zn and Te,respectively and the weight function for the incomplete charge collection described in 4.1. And we consider Compton scattering. Therefore our correction of spectra improves the accuracy. We modify the stripping formula presented by Castro et al. into the following formula; Emax N t (E 0 ) = {N d (E 0 ) - R(E 0, E) N t (E)} / ε(e 0 ) (4) E=E 0 Where N t (E 0 ): true number of photons of energy E 0, N d (E 0 ): number of photons detected of energy E 0, E max : maximum photon energy of the spectrum, R(E 0, E): monoenergetic response function of events of energy E due to interactions including incomplete charge collection, ε(e 0 ): full energy peak efficiency. The stripping procedure is applied step by step starting from the maximum photon energy E max. From the counts N d (E 0 ) of every energy E 0 is subtracted the monoenergetic response function R(E 0, E) in Fig. 5 of the true contents of each energy channel greater than E 0. The energy channel contents N t (E 0 ) so obtained are divided by the full energy peak efficiency ε(e 0 ) in Fig. 2. Consequently an example of the results is shown in Fig. 3. Fig. 3 shows the comparison of (a)measured and (b)corrected γ-ray spectra of an isotope 241 Am for Ge(solid line) and CZT(dotted broken line) detectors. The corrected spectrum for the CZT detector coincides with that for the Ge detector. RESULTS AND DISCUSSION The comparison of corrected diagnostic X-ray spectra at (a)70 kv and (b)100 kv for Ge and CZT detectors are shown in Fig. 6. The corrected spectra at (a)70 kv and (b)100 kv for the CZT detector coincide with those for the Ge detector. Each exposure attenuation curve for (a)70 kv and (b)100 kv is calculated and is compared with the curve measured by the ionization chamber in Fig. 7. The obtained results with the CZT detector agree with those obtained by the ionization chamber.

5 FIG. 6. Comparison of corrected diagnostic X-ray spectra at (a)70 kv and (b)100 kv for Ge and CZT detectors. FIG. 7. Comparison of exposure attenuation curves obtained from the CZT spectra and the ionization chamber at (a)70 kv and (b)100 kv. FIG. 8. Comparison of corrected mammographic(mo-mo) X-ray spectra at (a)25 kv and (b) 30kV for the CZT and Ge detectors. The comparison of corrected mammographic(mo-mo) X-ray spectra at (a)25 kv and (b)30 kv for the Ge and the CZT detectors are shown in Fig. 8. The corrected spectrum(a) at (a)25 kv and (b)30 kv for the CZT detector coincides with the spectral data(8) for the Ge detector published by Bureau of Radiological Health(BRH). Each exposure attenuation curve for (a)25 kv(mo-mo) and (b)28 kv(rh-rh) is calculated and is compared with the curve measured by the ionization chamber in Fig. 9. The obtained curves with the CZT detector agree with those obtained by the ionization chamber. The comparison of directly measured (at 120 kv, 3 ma) and corrected primary X-ray spectra of a CT scanner is shown in Fig. 10(a). In Fig. 10(a) the dotted line is the direct measured spectrum and the solid line is

6 the corrected spectrum. The corrected spectrum obtained with the CZT detector roughly agrees with CT X-ray spectral data(9) obtained using the Ge detector published by BRH shown in Fig. 10(b). The exposure attenuation curve at 120 kv and 3 ma is calculated and is compared with the curve measured by the ionization chamber in Fig. 11. The curve obtained with the CZT detector agrees with the curve obtained by the ionization chamber. FIG. 9. Comparison of exposure attenuation curves obtained from the CZT spectra and the ionization chamber at (a)25 kv(mo-mo) and (b)28 kv(rh-rh). FIG. 10. Comparison of (a)photon spectra measured and corrected with (b)ct X-ray spectral data(9) of BRH. FIG. 11. Comparison of exposure attenuation curves obtained from the CZT spectra and the ionization chamber at 120 kv and 3 ma. CONCLUSION Our purpose is to measure directly some diagnostic X-ray spectra using a low-efficiency CdZnTe(CZT)

7 detector developed recently and to find out the fittest response function of CZT detector to correct the measured spectra. Photon spectra produced by a single-phase 2-pulse diagnostic X-ray unit driven at 50 to 100 kv and a few ma, a mammographic X-ray unit at 25 to 32 kv and 3 times of 80 mas and an X-ray computed tomography (CT) scanner at 120 kv and 3 ma have been analyzed. Since detected spectra were distorted by the response of CZT detector and did not present the true photon spectra, the correction by the stripping procedure was applied. The response function of detector used in this procedure has been determined by an evaluation of interactions (Kescape, coherent scattering, Compton scattering) and incomplete charge collection calculated using the Monte Carlo method. The Monte Carlo simulations were continued with changing the important factors(mean path length of hole λ h, dead layer of CZT crystal and weight factor W q ) of incomplete charge collection until the fittest response function was found out. Corrected photon spectra were compared with the diagnostic X-ray spectral data published by Bureau of Radiological Health (BRH) measured using a Ge detector. Exposures were calculated from the corrected photon spectra and compared with exposures measured using an ionization chamber. These results obtained using CZT detector agreed with the diagnostic X-ray spectral data of BRH and exposures obtained by the ionization chamber. Therefore we think that the CdZnTe detector is able to use for quality control and quality assurance of some diagnostic X-ray units. REFERENCES 1. M.Matsumoto, H.Kanamori, T.Toragaito and A.Taniguchi, KEK Proceedings 96-4, Radiation Detectors and Their Uses, (1996) 2. M.Matsumoto,H.Kanamori, A.Taniguchi and M.Yoshida, Medical&Biological Engineering&Computing 35, Supplement Part 2 (1997) M.Matsumoto, H.Kubota, H.Kanamori and M.Yoshida, International Congress on Imaging Science (Antwerp,1998) Vol. 2 of Proceedings, W.W.Seelentag and W.Panzer, Phys. Med. Biol. 24, (1979) 5. C.S.Chen, K.Doi, C.Vyborny, H-P. Chan and G.Holje, Medical Physics 7(6), (1980). 6. C-T.Chen, H-P. Chan and K.Doi, Energy responses of germanium planar detectors used measurement of x-ray spectra in the energy range from 12 to 300keV:Monte Carlo simulation studies, Research Report UCHI-DR/84-01(Univ. Chicago 1984). 7. E.Di.Castro,R.Pani,R.Pellegrini and C.Bacci, Phys. Med. Biol. 29, (1984) 8. R.Fewell and R.E.Shuping, Handbook of mammographic X-ray spectra, Bureau of Radiological Health, Rockville, Maryland, U.S.A. (1978). 9. T.R.Fewell, R.E.Shuping and K.R.Hawkins,Jr., Handbook of computed tomography X-ray spectra, Bureau of Radiological Health, Rockville, Maryland, U.S.A. (1981).