Development of a Pixelated Detector for Clinical Positron and Single-photon Molecular Imaging

Transcription

1 Journal of Medical and Biological Engineering, 32(5): Development of a Pixelated Detector for Clinical Positron and Single-photon Molecular Imaging Hsin-Ching Liang 1,2 Meei-Ling Jan 1,* Jenn-Lung Su 2 1 Physics Division, Institute of Nuclear Energy Research, Longtan 325, Taiwan, ROC 2 Department of Biomedical Engineering, Chung-Yuan Christian University, Taoyuan 320, Taiwan, ROC Received 10 Jul 2011; Accepted 2 Oct 2011; doi: /jmbe.972 Abstract An imaging detector as a base unit for a gamma camera for dual-modality (positron and single-photon) molecular imaging is developed in this study. The imaging detector is constructed using scintillation material lutetium yttrium oxyorthosilicate (LYSO) and a position-sensitive photo-multiplier tube. Performance indices for dual-modality imaging applications are measured and the detector s feasibility is evaluated. Since LYSO exhibits spontaneous radiation, the practicality of the unit for single-photon imaging is evaluated. The results show that mean pixel widths of 0.6 and 0.8 mm and energy resolutions of 10% and 15% are achieved for 511 kev and 122 kev gammas, respectively, with a crystal pixel size of 1.5 mm. This pixel size is shown to be an engineering limit for meeting the requirement of the two imaging modalities. The background count rate from spontaneous radiation of 69 counts per second (cps) is less than 1/10 the size of the recorded signals in pinhole-collimated imaging. It is thus concluded that the proposed LYSO-based detector unit can be applied to dual-modality molecular imaging for clinical use with outstanding performance. Keywords: Pixelated imaging detector, Lutetium yttrium oxyorthosilicate (LYSO), Dual-modality imaging 1. Introduction Nuclear-medicine-based molecular imaging modalities have shown potential in the clinical fields [1-8], especially in mammographic applications [1-6]. The bio-functional information provided by these modalities has been shown to increase sensitivity and specificity for cancer diagnosis [7,8]. The development of a prototype breast scanner for imaging positron and single-photon tracers, such as 18 FDG and 99m Tc-MIBI, is planned in our laboratory. To construct the dedicated camera for this scanner, a gamma imaging detector which serves as a base unit and is capable of detecting annihilation and lower-energy gammas is required. The present study thus develops an imaging detector unit and measures its performance for the two imaging applications. The detector design is developed from positron imaging. It includes a pixelated crystal array and a thick scintillation layer. The pixel size is enlarged to fit single-photon imaging applications. Therefore, it is necessary to validate the feasibility and practicality of the proposed detector unit for single-photon imaging and to measure its imaging abilities for the two imaging applications. Dual-modality imaging units have been previously * Corresponding author: Meei-Ling Jan Tel: ext mljan@iner.gov.tw developed [9-11]. Due to the low density of yttrium aluminum perovskite (YAP), a thick (30 mm) crystal layer has been employed to improve gamma-capturing efficiency for positron imaging. However, this causes severe parallax error with increasing gamma incident angle. For a denser material such as gadolinium oxyorthosilicate (GSO), the thickness can be smaller. However, the poor light yield of GSO leads to a coarse pixel size, and thus poor imaging quality is expected. Therefore, a dense scintillation material with good light yield is required for this study. According to a literature review [12-15], among the several candidate materials, including GSO, lutetium gadolinium oxyorthosilicate (LGSO), bismuth germanate (BGO), yttrium oxyorthosilicate (YSO), YAP, lutetium oxyorthosilicate (LSO), and lutetium yttrium oxyorthosilicate (LYSO), LSO (Lu 2 (SiO 4 )O:Ce) exhibits the most suitable properties for gamma imaging. It has a fast signal (40 ns), a high light yield (75% of NaI(Tl)), and good gamma-capturing ability (11.6 mm attenuation length for 511 kev gammas). GSO has a cleavage problem, which makes it difficult to manipulate for assembling matrices for building imaging detector. LGSO does not have this drawback and has a better light yield (40% of NaI(Tl)). However, LGSO is inferior to LSO in terms of light output and gamma capturing. BGO is good at gamma capturing (11 mm for 511 kev gammas) but very poor in terms of light output and signal speed. YSO and YAP have good light output and fast signals but are poor at gamma capturing, degrading their usability for positron imaging, in particular YSO. LYSO is the

2 374 J. Med. Biol. Eng., Vol. 32 No only material comparable to LSO. LYSO exhibits similar properties and performance at relatively low cost compared to LSO which makes LYSO an acceptable alternative. The properties of LYSO lie between those of LSO and YSO with varying ingredient ratios of lutetium to yttrium [13]. LSO, YSO, and LYSO are all fast scintillators (about 40 ns). LYSO with a lutetium content of more than 30% shows light output very close to that of LSO. However, its gammacapturing ability degrades linearly with increasing yttrium content. Therefore, LYSO with only 5% of yttrium, i.e., Lu 1.9 Y 0.1 (SiO 4 )O:Ce, is used in this study due to its balanced scintillation properties. There are two major concerns when developing a LYSObased dual-modality imaging unit. The first concern is the gamma energy applied. Since the imaging physics of a gamma detector is a statistical process, the number of scintillation photons influences the signal-to-noise ratio and thus determines the resultant image quality. Since LYSO has almost the same relative light yields from 140 to 511 kev [13], the number of scintillation photons from a 140 kev gamma ray is only 27% of that from a 511 kev one. This implies that the image quality achieved with 140 kev gammas is poorer than that achieved with 511 kev gammas. Therefore, it is necessary to evaluate the performance at high and low gamma energies. Acceptable performances in both gamma energy bands is required for dual-modality imaging. The other concern is the physical properties of the scintillation material itself. There is spontaneous radiation originating from 176 Lu inside the scintillation crystal. According to previous studies [14,15], this radioactive isotope is an admixture in natural lutetium with an abundance of 2.6% and a half-life of years. 176 Lu undergoes β-decay (mean β-energy of 420 kev) and then emits gamma rays of 307 kev (94%), 202 kev (78%), and 88 kev (15%). The isotope results in a natural activity as background of about 300 cps/cm 3 for LSO. The background activity should be slightly lower for LYSO, since 5% of lutetium is replaced by yttrium, implying a specific activity of about 285 cps/cm 3. For the crystal array to be fabricated, a background activity of about 4000 cps is expected, which will seriously contaminate incoming gamma signals, especially in single-photon imaging applications. In positron imaging applications, because of a 511-keV energy window and a timing coincidence comparison serving as filters [16], the effects of the intrinsic activity are negligible. In contrast, in single-photon imaging, collimators must be applied, which physically suppresses sensitivity, and thus intrinsic radiation is expected to have a large effect. Therefore, it is necessary to evaluate the count rate performance and the feasibility of application to collimated single-photon imaging. 2. Materials and methods 2.1 Detector configuration The gamma imaging detector unit comprised a pixelated LYSO crystal array with dimensions of pixels. A single crystal pixel was mechanically cut and polished to a size of mm 3. Each pixel was covered with an adhesive aluminum film on its lateral surfaces for optical isolation and then assembled into an array block. The crystal block was optically coupled to a position-sensitive photo-multiplier tube (PSPMT, 8500 Hamamatsu) connected to a self-designed readout circuit. The assembled detector unit is shown in Fig. 1. The readout was a linear charge divider, which comprised several simple resister chains reducing 64 anode signals to 4 outputs. The purpose of this readout was to lower the rear-end electronics loading for signal conditioning and digitizing. Event position was determined using the following Anger logic calculation [17]: C D A B X A B C D A D B C Y A B C D where A, B, C, and D are digitized position signals from the readout circuit. The equations are simply algebraic relations among the four values of the signals. The calculated positions could be utilized to form a crystal map by the event position histogram, or they could be compared with a crystal look-up table (clut) for projection imaging. The effective imaging area of the assembled detector unit was mm 2 with a pitch of 1.7 mm between the centers of the crystal pixels. Figure 1. Photograph of the assembled crystal array (left). Detector unit formed by crystal array optically coupled to a PSPMT (right). 2.2 Evaluation methods Measurements of detector performance for high- and low-energy gammas and spontaneous background behavior were made. The data were analyzed to evaluate the practicality of the proposed detector unit for dual-modality imaging. To determine the influence of spontaneous radiation on singlephoton imaging, collimated single-photon imaging tests on phantoms were conducted. The images and count rate were analyzed. Details of the approaches are given below. The detector was exposed to 68 Ge (511 kev) and 57 Co (122 kev) flood sources to acquire data in various gamma energy bands. The acquired and digitized signals were substituted into Eq. (1) to be transformed into a 2D map showing the crystal matrix. This 2D map was rearranged to form a pixel flood image for further analysis. The flood images were first qualitatively (visually) examined to confirm that all the 625 crystal pixels were distinguishable. They were then processed with an in-house software tool, (1)

3 Development of Dual-modality Imaging Detector 375 which performs image processing, semi-automatically locates the peaks of event count distributions for crystal pixels, draws boundaries, builds and stores the clut, and finally calculates the energy spectrum for every pixel according to the clut. With the pixel profiles of 2D maps extracted for analysis, two indices of imaging performance, namely the pixel width (full width at half maximum (FWHM) of the profile) and the peakto-valley ratio, were obtained. These two indices represent the pixel resolution of the detector unit in imaging. A narrower pixel width and a higher peak-to-valley ratio mean clearer pixel distinction and less overlap of event counts, respectively, corresponding to better pixel resolution and image quality. By analyzing the pixelated energy spectrum, two more indices of imaging performance, namely the energy resolution and the uniformity over the whole array, were obtained. The four indices represent the performance of the detector unit in real imaging applications. To evaluate the influence of natural radiation from lutetium, two-hour background measurements without any radioactive source were made three times. The energy spectrum and count rate were then analyzed. 2.3 Single-photon imaging tests Since the detector is mainly designed for positron imaging, besides measurement of its imaging performance indices, its feasibility of application to practical single-photon imaging was validated. To achieve this, a collimated imaging platform, which holds the detector, collimator, and phantom together and aligns them on a linear stage, was built. A pinhole collimator or a parallel-hole collimator were used according to the requirement of the tests. The pinhole was made by tungsten with a 1-mm aperture, and the parallel holes (2 mm in diameter) were formed by thin lead septa. The imaging platform is shown in Fig. 2. The raw data acquired were processed by comparison with the clut, energy windowing (140 kev ± 10%), and uniformity correction (by an efficiency correction table derived from a 140-keV 2D map/flood image) to form a pixel projection image, which was the result of single-photon collimated imaging. inner diameter of 1.0 mm and filled with 1.46 mci (0.7 and 0.76 mci) 99m Tc were used for the phantom studies. Three images were acquired as follows. (1) Two thin line sources 1 mm apart were imaged at 13 mm from the pinhole and 600k counts were acquired. (2) Two thick line sources 2.5 mm apart were imaged at 33 mm from the pinhole and 300k counts were acquired. (3) Two thick line sources crossing each other were imaged at 33 mm from the pinhole and a 1-minute imaging was made to examine the image quality under practical single-photon imaging conditions. In each measurement, the acquisition time was recorded for evaluating the count rate performance. For the parallel-hole imaging test, an acrylic phantom with seven pipes was used. Each pipe was 2 mm wide. Details of the phantom are depicted in Fig. 3. The phantom was filled with 2.7 mci 99m Tc and positioned at 65 mm from the collimator face for scanning. 2 units: mm Figure 3. Photograph of the 7-pipe phantom. Each pipe is 2 mm wide and 2 mm deep. The marked dimensions have units of millimeters. For all measurements, the imaging results were visually examined and the count rate versus background behavior was compared. Before the images were acquired, a flood phantom filled with 2 mci 99m Tc was imaged at 33 mm (pinhole) and 130 mm (parallel-hole) from the collimator. In both cases, two 10M-count acquisitions were made. These flood images were used for establishing the clut and the efficiency (or uniformity) correction table Results and discussion 3.1 Performance index measurements Figure 2. Photograph of the installed collimated imaging platform. The white portion in is a pinhole collimator which can be replaced by a parallel-hole collimator depending on imaging demand. For the pinhole collimation test, two thin capillaries with an inner diameter of 0.5 mm and filled with 1.73 mci (0.78 and 0.95 mci) 99m Tc (140 kev), and two thick capillaries with an The resultant flood images relative to 68 Ge and 57 Co are shown in Fig. 4. Visually separable pixels, totaling 625 (a array), were observed in both images. The analyzed performance indices are listed in Table 1. Good capabilities for imaging both high- and low-energy gammas are shown.

4 376 J. Med. Biol. Eng., Vol. 32. No Pixel FWHM: 0.6 mm Peak-to-valley ratio: 7.6 Eng. resolution: 9.8% Uniformity: 0.59 (a) (b) Pixel FWHM: 0.8 mm Peak-to-valley ratio: 5.0 Eng. resolution: 15% Uniformity: 0.56 Figure 4. (a) Flood images from exposure of 68 Ge (left) and 57 Co (right). (b) In each image, all 625 crystal pixels can be distinguished and separated for building the clut. Although both images show a distinguishable crystal matrix, the one from 57 Co is obviously more ambiguous than that from 68 Ge (Fig. 4(a)). The mean pixel width (FWHM) and energy resolution in Table 1 also show a consistent phenomenon, which can be attributed to a decrease in the number of scintillation photons under lower gamma energy. Since the relative light yield of LYSO is almost constant relative to the gamma energy between 100 kev to 1 MeV [18], the number of emission photons under 57 Co exposure was only about 23% of that under 68 Ge exposure. Such a difference might increase statistical error which degrades the quality of 2D maps and the resultant images. In addition, the reduction in crystal size that could be distinguished under this low-energy gamma exposure is also limited. In the test on pixel size selection conducted prior to this study, a successfully separated image of an array composed of mm 3 segments under 511-keV exposure was obtained. However, when the 122-keV gamma source was used, small pixels, including those in an array composed of mm 3 crystals, were no longer resolvable. The estimated separable limitation is about mm 2 under 122-keV gamma exposure. Therefore, mm 3 crystal pixels were chosen for further assembly of the scintillation block. There is a trade-off between detector performance and applicability for low-energy gammas. According to the attenuation lengths of scintillators for high- and low-energy gammas in [19], reasonable estimations for those at 140 kev and 511 kev of LYSO are 1~1.5 mm and 12 mm, respectively. To image high- and low-energy gammas, a 10-mm crystal depth was chosen for capturing 511-keV gammas. Compared to the suitable crystal depth for single-photon imaging, i.e., the attenuation length at 140 kev, the chosen 10 mm is much higher. Due to the self-absorption of the scintillation photons within the LYSO material [14,15,18,19], a longer crystal means a longer movement path before the photons get out, thus reducing scintillation light output. Therefore, the proposed detector is expected to have degraded single-photon imaging and 2D maps under lowenergy gamma exposure due to the 10-mm crystal depth, which is a compromise for dual-modality imaging. The measured performance indices listed in Table 1 show that the proposed detector has good imaging abilities for both high- and low-energy gammas. Compared with the spatial performances reported in [11], i.e., peak-to-valley ratios of 12 and 4, and pixel FWHM values of 0.5 mm and 0.8 mm for 511- and 122-keV gammas, respectively, those of the proposed detector unit are worse (0.64 vs. 0.5 mm) under 511-keV gamma exposure. Besides the differences in detector configuration, the data in [11] were acquired in coincidence mode, whereas the data in this work were obtained in single mode. Coincidence-mode data acquisition keeps signals from non-annihilation radiation and filters out noise, especially that from the spontaneous background, resulting in better images acquired and more accurate spatial performance index. For performance under 122-keV exposure, the results obtained here are slightly better than those in [11]. However, the differences are within the standard deviation, so the performances should be viewed as being comparable. Table 1. Perfomance indices for two gamma energy bands Gamma energy (kev) Pixel FWHM (mm) 0.64 ± ± 0.22 Peak-to-valley ratio 7.64 ± ± 2.8 Energy resolution 9.8% ± 2.2% 15% ± 1.5% Uniformity 0.59 ± ± 0.09 In gamma-ray imaging, the crystal pixel size directly affects the resolution of the imaging system. Size reduction is restricted by the light yield of the scintillator. For example, GSO (light yield 20% of NaI(Tl)) was chosen in a previous study for building a detector for both positron and singlephoton imaging [10]. Owing to the lower light output and the use of low-energy gammas, a coarser size of mm 2 was selected. Therefore, a resolution of below 3 mm for 511-keV gammas was reported. To improve resolution or reduce pixel size, especially for single-photon imaging, replacing LYSO with materials with higher light yield should be a viable solution. NaI(Tl) and some newly developed scintillation materials [19] are potential candidates. Compared with LSO and LYSO, these materials have better light output and no intrinsic radiation. A higher signal-to-noise ratio and lower statistical error would lead to better image quality. However, these materials have inferior gamma-stopping ability [19,20], which could degrade detection efficiency (or detection sensitivity) when applied to positron imaging. To compensate, increasing the thickness of the scintillator layer was necessary [9,11]. However, this increases the possibility of gamma rays penetrating through one pixel to another (i.e., parallax error). When lighter and longer crystals were applied, the image

5 Counts Count s Development of Dual-modality Imaging Detector 377 quality decreased due to parallax error, especially in positron imaging applications. To avoid such effect, crystal pixels of a larger size is suggested [10] but this would degrade the resolution. Another solution is to develop imaging detectors which can offer depth-of-interaction information at higher cost. With these engineering factors considered, LYSO doped with 5% yttrium was a best fit available to meet the requirements of the proposed study. 3.2 Background behavior In long-term background measurements, the average count rate was 1714 cps. The energy spectrum shown in Fig. 5 is consistent with that given in [14]. By taking the total volume of the scintillation material (625 pixels of mm 3 crystals, for a volume of cm 3 ) into consideration, a specific count rate of 122 cps/cm 3 was obtained. When an energy window of 140 kev ± 10% was applied, the specific background became 5 cps/cm 3 (average count rate of 69 cps). When a 122-keV ± 10% energy window was applied, the recorded background count rate was almost the same x 10 4 Energy spectrum of lutetium natural background 122 kev 511 kev radiation energy (kev) Figure 5. Energy spectrum of background measurement of the LYSObased detector without any radioactive source. The specific count rate of 122 cps/cm 3 is much lower than that of 263 cps/cm 3 reported in [18]. Such a difference is mainly attributed to the large-area crystal array coupled to the large-area PMT. The light-conducting area from crystal to the PMT in [18] was only mm 2, whereas the crystal block used here covered an area of mm 2. Moreover, the PMT used in this study had a large effective area of mm 2. These increased the dead-time loss of the detector itself. Therefore, the reduction in the recorded counts is reasonable and as expected. Another background measurement was made with a smaller block, an array composed of crystals of the same size (covering an area of mm 2 ), coupled to the same PMT. A count rate of 558 cps was obtained, implying a specific count rate of 205 cps/cm 3, which is closer to that in [18]. 3.3 Single-photon imaging tests Images obtained from collimated projection tests are shown in Fig. 6 to Fig. 8. Figure 6 shows an image from acquisition 1 of the pinhole imaging tests, and Fig. 7 displays those from acquisitions 2 and 3. One of the results of the parallel-hole imaging tests is shown in Fig. 8. The count rate performances for all tests are listed in Table 2. All images shown in Fig. 6 to Fig. 8 show meaningful results compared with their imaging objects, implying that the 0.9mm Figure 6. Pinhole-collimated projection image (right) of 2 thin capillaries of 99m Tc source positioned 13 mm in front of the pinhole. The effective activity was mci and the center pitch of the 2 capillaries was 2 mm. A profile of center cross-section (left) showing the separation of 2 thin line sources, implying that the device could offer good imaging quality even under single-photon collimated conditions Figure 7. Pinhole-collimated projection images of 2 thick capillaries of 99m Tc source positioned 33 mm in front of the pinhole, standing 2.5 mm apart (left) and crossing each other (right). The effective activity was 0.29 mci. Both images show promising quality with respect to physical phantom arrangement. The image on right was a one-minute acquisition image, indicating the practicability of applying the detector to single-photon imaging Figure 8. Parallel-hole-collimated projection images of 7-pipe phantom (refer to Fig. 3, filled with 99m Tc) located 65 mm in front of collimator. The effective activity was about 2.3 mci. Image resolution is much coarser than that obtained with pinhole collimation.

6 378 J. Med. Biol. Eng., Vol. 32 No Table 2. Count rate performances for collimated imaging tests. Applied collimator Condition of measurement Background Pinhole Parallel-hole Phantom -- 2 thin 2 thick lines lines 7-pipe Filled activity (mci) Effective imaged activity (mci) Source-collimator distance (mm) Avg. count-rate (cps) count rate was statistically high enough to cover the spontaneous background from lutetium. As shown in Table 2, the recorded count rate was up to 10 times higher than that from the background, even under 1-mm pinhole collimation. With source activity taken into account, the two thin capillaries were filled sources to a length of 50 mm (0.78 mci) and 60 mm (0.95 mci). For a pinhole collimator with a 90-mm focal length, the actual effective imaging activity was only mci relative to the imaging condition with a pinhole-source distance of 13 mm. For a 1-mm pinhole and an object 13 mm away, an activity of mci resulted in an average count rate of about 830 cps. Assuming that the lowest signal-to-background ratio that could lead to an acceptable image quality is 3, i.e., 2 times the signals of the backgrounds, a recorded count rate of 207 cps can be an acceptable lower limit. In other words, the lowest activity that could offer a meaningful image is about 0.05 mci within a mm 2 area. In parallel-hole imaging, the actual effective imaging activity was 2.3 mci, yielding a count rate of about 1300 cps. Therefore, the lowest estimated acceptable activity is about 0.36 mci within an imaging area of mm 2. These estimations imply that the proposed detector unit can be applied to single-photon in-vivo imaging. It should be noted that energy values were windowed and screened in the software during image acquisition, which would reduce the recorded count rate. With dedicated electronics applied, much higher count rates might be achieved. 4. Conclusion A pixelated gamma imaging detector unit was developed and its performance indices for imaging positrons and single photons were measured. The feasibility of using the detector for single-photon imaging was proven. According to the performance results, the proposed imaging detector unit shows good imaging ability for both high and low gamma energy bands. It not only meets the primary demand for positron imaging, but also shows potential for single-photon imaging applications. Even under the influence of the 176 Lu natural background, a meaningful count rate in collimated conditions was obtained. Hence, the proposed LYSO-based detector unit has sufficient capability for dual-modality imaging applications. Although the scintillator itself exhibits a specific count rate of several hundred cps with a suitable energy window applied (140 kev ± 10% in this study), a truly effective background noise leaves only 5 cps/cm 3. For the proposed detector unit, the recorded count rate of the intrinsic background is 69 cps, which is less than 1/10 the signal count rate under collimated conditions, indicating that it is practicable for application to single-photon imaging. For conditions with collimators, the number of useful gamma rays reaching the detector is severely reduced. With a low effective background, count rates high enough to offer meaningful images can be achieved. When building a clinical imaging device in the future, for example a nuclear-medicine-based breast imaging camera, an imaging head with a larger and consecutive effective area is required. Continuous light guides coupled to the multi-pspmt array can be used to extend the imaging area. When applying a light guide as an inter-medium between the scintillator and the PMT window, refractions and reflections at the interfaces and absorptions inside the light guide of the scintillation photons cause losses, thereby reducing the number of photons that reach the PMT [21]. Such reduction degrades detector performance, which becomes even worse under low-energy gamma exposure. Therefore, crystals of a larger size should be used, especially when low-energy gammas are applied. For the spontaneous background, the situation will not get worse since each PMT shares approximately the same volume of the scintillator as the unit depicted here. A similar level of signal-to-background count rate for a larger-area imaging head is expected. Acknowledgments The author would like to acknowledge Mr. K. W. Chen for readout board fabrication and testing, and Mr. C. H. Yeh for acquiring the radioactive source 99m Tc. References [1] F. Giammarile and A. Bremond, Diagnostic of breast cancer: what do clinicians expect from PEM?, Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip., 527: 83-86, [2] W. W. Moses, Positron emission mammography imaging, Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip., 525: , [3] P. Rodrigues, R. Moura, C. Ortigão, L. Peralta, M. G. Pia, A. Trindade and J. Varela, Geant4 applications and developments for medical physics experiments, IEEE Trans. Nucl. Sci., 51: , [4] J. Qi, C. Kuo, R. H. Huesman, G. J. Klein, W. W. Moses and B. W. Reutter, Comparison of rectangular and dual-planar positron emission mammography scanners, IEEE Trans. Nucl. Sci., 49: , [5] D. P. McElroy, E. J. Hoffman, L. MacDonald, B. E. Patt, J. S. Iwanczyk, Y. Yamaguchi and C. S. Levin, Evaluation of breast tumor detectability with two dedicated compact scintillation cameras, IEEE Trans. Nucl. Sci., 49: , [6] M-L. Jan, K-S. Chuang, Y-C. Ni, C-C. Pei, J. Wu, C-K. Yeh and Y-K Fu, Feasibility study of using PEImager scanner for positron emission mammography, IEEE Trans. Nucl. Sci., 52: , [7] B. Bagni, A. Franceschetto, A. Casolo, M. De Santis, I. Bagni, F. Pansini and C. Di Leo, Scinti-mammography with 99mTc- MIBI and magnetic resonance imaging in the evaluation of breast cancer, Eur. J. Nucl. Med. Mol. Imaging, 30: , [8] O. Schillaci and J. R. Buscombe, Breast scintigraphy today: indications and limitations, Eur. J. Nucl. Med. Mol. Imaging, 31: S35-S45, 2004.

7 Development of Dual-modality Imaging Detector 379 [9] C. Daminani, A. Del Guerra, G. Di Domenico, M. Gambaccini, A. Motta, N. Sabba and G. Zavattini, An integrated PET-SPECT imager for small animals, Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip., 461: , [10] S. Yamamoto, K. Matsumoto and M. Senda, Development of a GSO positron/single-photon imaging detector, Phys. Med. Biol., 51: , [11] D. J. Herbert, N. Belcari, M. Camarda, A. D. Guerra and A. Vaiano, Characterisation of crystal matrices and single pixels for nuclear medicine applications, Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip., 537: , [12] S. Shimizu, K. Kurashige, T. Usui, N. Shimura, K. Sumiya, N. Senguttuvan, A. Gunji, M. Kamada, and H. Ishibashi, Scintillation properties of Lu 0.4Gd 1.6SiO 5:Ce (LGSO) crystal, IEEE Trans. Nucl. Sci., 53: 14-17, [13] T. Kimble, M. Chou and B. Chai, Scintillation properties of LYSO crystals, Proc. IEEE NSS/MIC, 3: , [14] T. Ludziejewski, K. Moszynska, M. Moszynski and D. Wolski, Advantages and limitations of LSO scintillator in nuclear physics experiments, IEEE Trans. Nucl. Sci., 42: , [15] K. S. Shah., P. Bennett and M. R. Squillante, Gamma ray detection properties of lutetium aluminate scintillators, IEEE Trans. Nucl. Sci., 43: , [16] A. P. Dhawan, Medical image analysis, IEEE Press on Biomedical Engineering, Ch.4, [17] S. Siegel, R.W. Silverman, Y. Shao and S. R. Cherry, Simple charge division readouts for imaging scintillator arrays using a multi-channel PMT, IEEE Trans. Nucl. Sci., 43: , [18] L. Pidol, A. Kahn-Harari, B. Viana, E. Virey, B. Ferrand, P. Dorenbos, J. T. M. de Haas and C. W. E. van Eijk, High efficiency of lutetium silicate scintillators, Ce-doped LPS, and LYSO crystals, IEEE Trans. Nucl. Sci., 51: , [19] W. W. Moses and K. S. Shah, Potential for RbGd 2Br 7:Ce, LaBr 3:Ce, LaBr 3:Ce, and LuI 3:Ce in nuclear medical imaging, Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip., 537: , [20] G. F. Knoll, Radition Detection and Measurement, 3rd (Ed), John Wiley and Sons Inc., Ch.8, [21] H-C. Liang, M-L. Jan, J-L. Su, W-C. Lin, S-F. Yu and L-H. Shen, Development of an LYSO based gamma camera for positron and scinti-mammography, J. Instrum., 4: P08009, 2009.

8 380 J. Med. Biol. Eng., Vol. 32 No