- Email: [email protected]
CZT gamma camera for scintimammography
2004 Workshop on the Nuclear Radiology of Breast Cancer Rome (Italy) October 22-23, 2004
((Physica Medical) 9Vol. XXI, Supplement l, 2006
CZT Gamma Camera for Scintimammography Ira M. Blevis, 1 M. K. O ' C o n n o r , z Z. Keidar, 3 A. Pansky, 1 H. Altman,1J. W. H u g g 1 1. GE Healthcare, Haifa, Israel 2. Mayo Clinic, Rochester, Minnesota, USA 3. Rambam Medical Center, Haifa, Israel Abstract
A high performance prototype gamma camera based on the semiconductor radiation detector Cd(Zn)Te is described. The camera features high spatial resolution, high-energy resolution, a reduced dead space on the edge of the field of view, and a compact format. The camera performance was first examined by comparison of small field of view examinations with those from an Elscint SP6HR standard clinical gamma camera. The new camera was found to give equal or improved image quality. The camera was then used for a systematic phantom study of small lesions in a background as would be found in breast cancer imaging. In this study the camera was able to systematically detect smaller, deeper, and fainter lesions. The camera is presently being used in a clinical trial aimed to assess its value in scintimammography where previous limitations of image quality and detector size have restricted the use of the functional imaging techniques. Preliminary results from 40 patients show high sensitivity and specificity with respect to X-ray mammography and surgery. KEYWORDS:CTZ, scintimammography; Gamma Camera, breast. 1. INTRODUCTION Medical I m a g i n g plays an i m p o r t a n t role in the diagnosis, treatment, and prevention of breast cancer. T h e m a i n modality in current use is X-ray m a m m o g raphy that gives high-resolution images of anatomical structures that block X-rays. Since breast cancer (BC) grows f r o m a small initial phase that m a y be structurally similar to healthy tissue, there is some difficulty in b o t h detecting all BC in the population and also in positively identifying detected features as BC. T h e n o m i n a l Sensitivity for X-ray m a m m o g raphy of 90% is reduced in the population due to 20% occurrence of dense breast tissue and also 20% occurrence of undetectable lobular cancer. T h e specificity (with respect to biopsy) is only 25% resulting in increased anxiety levels and further costs. T h e efficacy of the accepted medical practice including X - m a m m o g r a p h y screening in the general population is q u o t e d as 30% reduction in BC [1]. Other i m a g i n g modalities have been considered for reducing the gaps left by X-ray m a m m o g r a p h y . Ultrasound is also useful for the visualization of structure, but has the added limitations of lower contrast and a strong dependence on operator skills. Recent advances in MRI have extended its use to angiogenesis indicators (Gd), and yielded improved specificity values of -80%; however, the cost of the technique precludes its use in screening at this time. Functional i m a g i n g with 99mTc and g a m m a cameras has long been a candidate for resolving the problems of clinical performance and cost. Indeed s c i n t i m a m m o g r a p h y is used occasionally, but the general use has been limited by difficulties in finding small (i.e. young) lesions. Although high-resolution systems can be built to ameliorate the problem, the
size of the camera heads required by the standard scintillator and photomultiplier tube t e c h n o l o g y prevents optimal viewing and limits the solution. Specific problems include the visualization of medial lesions, the large c o m p o n e n t of scatter radiation f r o m the myocardium, and the inability to localize the lesions correctly in the anatomy. Overall, the detectability of lesions < 1 c m has been reported to be < 50% [2]. It has been since suggested that the limitations of the NaI g a m m a cameras have contributed to the unrealized potential of scintim a m m o g r a p h y . PET imaging m a y be able to give the n e e d e d i m p r o v e m e n t in image quality, b u t like MRI it is not suitable for screening because of the present levels of expense. C Z T - s c i n t i m a m m o g r a p h y has recently been developed [3,4] and may prove to be the solution to the problems. C Z T - s c i n t i m a m m o g r a p h y is based on the direct conversion of g a m m a energy to electric charge w i t h o u t the intermediary steps of scintillation production and detection. One i m p o r t a n t advantage that ensues is that the detector material, electronics, and total package need only be a few centimeters thick and with no extra material a r o u n d the imaging margins. Thus it can be used for m a m m o g r a p h y - l i k e viewing right up to the chest wall w i t h o u t including a b a c k g r o u n d signal f r o m the chest or myocardium. Another key advantage is the direct conversion of the g a m m a energy to a confined locus of electric charge in the semiconductor that in t u r n gives b o t h g o o d energy resolution and a significantly improved intrinsic spatial resolution. T h e combination of high fidelity functional imaging and close approach to the subject allow significantly better small feature discovery and identification than has been realized by other practical techniques.
Address for correspondence: Ira M. Blevis, GE Healtheare, Keren Hayesod 16, building 5, Tirat Hacarmel, Haifa. E-mail: ira. [email protected] 3. N e w C a m e r a I n s t r u m e n t a t i o n
<
2. MATEPIALS AND METHODS
A test detector was built in the physics laboratory at Gene:al Electric Healthcare, Haifa. It was not specifically intended for CZT-scintimammography. The dete :tot had an imaging surface of 20cmx20cm compri:;c d of 6400 2.5mmx2.5mm pixels. It weighed 10 kg (w t h o u t the collimator) and was 8 cm thick. The ene::gy resolution was 6.5% fwhm for 99mTc. Correction maps included energy maps and sensitivity ir aps, but not linearity maps. It was equipped with GP and HR collimators that had holes aligned with the C Z T pixels. The Xpert software package for Nuc] ear Medicine acquisition and display was adapted I:o accept the pixelated data just as continuous Nal data. After completing laboratory testing for functionality arLc stability, initial performance assessment was petf:)rmed using side by side imaging of clinical subject:; at R a m b a m Medical Center, Haifa. Patients undergoing bone studies on an SP6HR, a standard high resolution (95 pmt's) NaI g a m m a camera equipped with a HR collimator, were imaged with the sarrLc view and same acquisition times ( - 5 minutes) on the C Z T system a few minuets later. Scint~inammography performance was tested at the Mayo Clinic, MN, using a water p h a n t o m and a set of small tillable balls of different sizes, immersed -o different levels, and in varying depths of wat,r:. The parameter space is shown in Table 1. Acqui:dtions using the p h a n t o m were made side by side with the CZT camera and a Helix g a m m a
camera to compare lesion detectability over the entire parameter space. Preliminary investigations of clinical scintimammography were also carried out at the Mayo Clinic. A direct comparison image on the CZT camera for a patient undergoing scintimammography was acquired. In addition a clinical trail involving 100 patients referred to CZT-scintimammography from X-ray m a m m o g r a p h y was commenced. Referral included patients with positive m a m m o g r a p h y (grade 4-5) and detected lesions of size 5-20 m m . Images were acquired using X-ray m a m m o g r a p h i c imaging positions. For each patient, cranio-caudal and/ or medio-lateral oblique views of 10 minutes each were acquired of each breast, with light (pain free) compression. 3. RESULTS
Figure 1 shows bone extremity images acquired with the C Z T camera and the SP6HR under the same conditions. The CZT images permit superior visualization of various features, for example the articular surfaces of the knee, the definition of the patella, and the metatarsal bones of the foot. The CZT images begin to have the character of X-rays images. The breast p h a n t o m lesion detectability p h a n t o m was imaged with the CZT and the Helix systems. The systems had similar sensitivities, therefore the acquisitions times and counts were also close. The main differences visible in the images are attributable
TAB. 1. Phantom Variables.
Phantom
Scintimammography
Ball diameter Water depth Depth of balls lAbj : [awater] Acquisition time Collimator
Size of tumor Tissue thickness Tumor location Tumor drug uptake Count density Collimator
Range
Units
5-7
cm cm
1-7 3:1-12:1 0.9, 2.2 GP / HR
Mcnts/ 400cm2
FIG. 1. Clinical comparison of CZT and SP6HR. 57
Ira M. Blevis et
alii:
CZT Gamma Camera for Scintimammography
to differences in the system resolution derived from the pixelation and collimator combination. Figure 2 shows a comparison of images for three values of the depths of the balls. In all cases the CZT shows higher clarity. This clarity is highlighted in the third comparison pair where the 4.2 m m ball and the 4.9 m m ball disappear from the NaI image. In randomized observer trails based on these images a systematic improvement in the n u m b e r of balls and thereby the (small) limit of detectability was shown by the CZT camera, particularly for balls 5-7 m m in diameter. The first clinical CZT-scintimammography comparison is shown in Figure 3. In this case the patient was being checked for reemergence of the primary t u m o r a year after surgical removal. The US and MRI were inconclusive due to the previous tissue damage; the NaI alone would have failed to detect the uptake of 99mTc;however, the CZT image taught the correct interpretation of the NaI image. The clinical trial of CZT-scintimammography has examined 80 of the 100 patients intended. At the time of writing the results were compiled for 40 of the studies. The 40 patients included 36 tumors. Twentysix patients had a total of 36 lesions confirmed at surgery Of these 36 lesions, 33 were detected by the C Z T system (overall sensitivity 92%). There were 22 lesions < 10 m m in diameter (Avg. 8.1 + 1.9 ram), and 19 were detected on C Z T imaging (sensitivity 86%). Two patients had false negative studies. In these pa-
FIG. 2. Breast Lesion Detectability Phantom Images. 58
tients, three lesions that were located deep within the breast were missed primarily due to (future avoidable) technical errors in positioning. M1 14 lesions > 1 cm in diameter seen on the C Z T system were confirmed at surgery. In three patients, lesions were detected by the CZT system in the contralateral breast to the breast containing the suspicious lesion seen on mammography. These lesions were seen on MRI and all were confirmed as true positive cases at surgery A total of 14 patients had negative findings at biopsy or surgery. Of these 9 were true negative and 5 were false positive on the CZT. False positive studies were benign fibroadenoma (3 patients), inflammatory fatty necrosis (1 patient), and benign breast parenchyma (1 patient). Figure 4 shows an example of the image quality realized in these studies. 4. CONCLUSIONS
A promising solution to the problems of dense breast tissue and possibly false positive studies in Xray m a m m o g r a p h y has been proposed in the form of CZT-Scintimammography. Preliminary investigations have been completed. The image quality for close objects is superior to that of NaI cameras in a variety of tests. The sensitivity and specificity is high from these early results. Small lesions in breast cancer < 1 cm should be reliably detected by the technique. Complete results will be available with the finish of the first clinical trial.
~Physica Medica~ - Vol. XXI, Supplement 1, 2006
Rt. b r e a s t - I D C 5 mm diameter
Lateral
Medial
Fro. 4. CZT-scintimammography: Clinical Trial. FIG, 3. CZT-scintimammography Comparison. R E F E R E N CES
[1] [2]
Gra owJ. Breast Cancer 2004, Progress and Promise on the :21inical Front, In this Issue. Kha Ikhali I, Villanueva-Meyer J, Edell SL, et alii Diagnostic ~ :curacy of 99mTc-sestaMIBI breast imaging: multiceiu :r trial results. J Nucl Med, 2000: 41; 1973-1979.
[3] [4]
Blevis I, Reznik A, Pansky A, Liss J, Wainer N. Small Field of View CZT Gamma Camera, IEEE Medical Imaging Conference, 2002. Mueller B, O'Connor M K, Blevis I, Rhodes D J, Smith R, Collins D A, PhillipsJ. Evaluation of a Cadmium Zinc Telluride Detector for Scintimammography, J Nuc Med, 2003: 44; 602-609
59
((Physica Medical) 9Vol. XXI, Supplement l, 2006
CZT Gamma Camera for Scintimammography Ira M. Blevis, 1 M. K. O ' C o n n o r , z Z. Keidar, 3 A. Pansky, 1 H. Altman,1J. W. H u g g 1 1. GE Healthcare, Haifa, Israel 2. Mayo Clinic, Rochester, Minnesota, USA 3. Rambam Medical Center, Haifa, Israel Abstract
A high performance prototype gamma camera based on the semiconductor radiation detector Cd(Zn)Te is described. The camera features high spatial resolution, high-energy resolution, a reduced dead space on the edge of the field of view, and a compact format. The camera performance was first examined by comparison of small field of view examinations with those from an Elscint SP6HR standard clinical gamma camera. The new camera was found to give equal or improved image quality. The camera was then used for a systematic phantom study of small lesions in a background as would be found in breast cancer imaging. In this study the camera was able to systematically detect smaller, deeper, and fainter lesions. The camera is presently being used in a clinical trial aimed to assess its value in scintimammography where previous limitations of image quality and detector size have restricted the use of the functional imaging techniques. Preliminary results from 40 patients show high sensitivity and specificity with respect to X-ray mammography and surgery. KEYWORDS:CTZ, scintimammography; Gamma Camera, breast. 1. INTRODUCTION Medical I m a g i n g plays an i m p o r t a n t role in the diagnosis, treatment, and prevention of breast cancer. T h e m a i n modality in current use is X-ray m a m m o g raphy that gives high-resolution images of anatomical structures that block X-rays. Since breast cancer (BC) grows f r o m a small initial phase that m a y be structurally similar to healthy tissue, there is some difficulty in b o t h detecting all BC in the population and also in positively identifying detected features as BC. T h e n o m i n a l Sensitivity for X-ray m a m m o g raphy of 90% is reduced in the population due to 20% occurrence of dense breast tissue and also 20% occurrence of undetectable lobular cancer. T h e specificity (with respect to biopsy) is only 25% resulting in increased anxiety levels and further costs. T h e efficacy of the accepted medical practice including X - m a m m o g r a p h y screening in the general population is q u o t e d as 30% reduction in BC [1]. Other i m a g i n g modalities have been considered for reducing the gaps left by X-ray m a m m o g r a p h y . Ultrasound is also useful for the visualization of structure, but has the added limitations of lower contrast and a strong dependence on operator skills. Recent advances in MRI have extended its use to angiogenesis indicators (Gd), and yielded improved specificity values of -80%; however, the cost of the technique precludes its use in screening at this time. Functional i m a g i n g with 99mTc and g a m m a cameras has long been a candidate for resolving the problems of clinical performance and cost. Indeed s c i n t i m a m m o g r a p h y is used occasionally, but the general use has been limited by difficulties in finding small (i.e. young) lesions. Although high-resolution systems can be built to ameliorate the problem, the
size of the camera heads required by the standard scintillator and photomultiplier tube t e c h n o l o g y prevents optimal viewing and limits the solution. Specific problems include the visualization of medial lesions, the large c o m p o n e n t of scatter radiation f r o m the myocardium, and the inability to localize the lesions correctly in the anatomy. Overall, the detectability of lesions < 1 c m has been reported to be < 50% [2]. It has been since suggested that the limitations of the NaI g a m m a cameras have contributed to the unrealized potential of scintim a m m o g r a p h y . PET imaging m a y be able to give the n e e d e d i m p r o v e m e n t in image quality, b u t like MRI it is not suitable for screening because of the present levels of expense. C Z T - s c i n t i m a m m o g r a p h y has recently been developed [3,4] and may prove to be the solution to the problems. C Z T - s c i n t i m a m m o g r a p h y is based on the direct conversion of g a m m a energy to electric charge w i t h o u t the intermediary steps of scintillation production and detection. One i m p o r t a n t advantage that ensues is that the detector material, electronics, and total package need only be a few centimeters thick and with no extra material a r o u n d the imaging margins. Thus it can be used for m a m m o g r a p h y - l i k e viewing right up to the chest wall w i t h o u t including a b a c k g r o u n d signal f r o m the chest or myocardium. Another key advantage is the direct conversion of the g a m m a energy to a confined locus of electric charge in the semiconductor that in t u r n gives b o t h g o o d energy resolution and a significantly improved intrinsic spatial resolution. T h e combination of high fidelity functional imaging and close approach to the subject allow significantly better small feature discovery and identification than has been realized by other practical techniques.
Address for correspondence: Ira M. Blevis, GE Healtheare, Keren Hayesod 16, building 5, Tirat Hacarmel, Haifa. E-mail: ira. [email protected] 3. N e w C a m e r a I n s t r u m e n t a t i o n
<
2. MATEPIALS AND METHODS
A test detector was built in the physics laboratory at Gene:al Electric Healthcare, Haifa. It was not specifically intended for CZT-scintimammography. The dete :tot had an imaging surface of 20cmx20cm compri:;c d of 6400 2.5mmx2.5mm pixels. It weighed 10 kg (w t h o u t the collimator) and was 8 cm thick. The ene::gy resolution was 6.5% fwhm for 99mTc. Correction maps included energy maps and sensitivity ir aps, but not linearity maps. It was equipped with GP and HR collimators that had holes aligned with the C Z T pixels. The Xpert software package for Nuc] ear Medicine acquisition and display was adapted I:o accept the pixelated data just as continuous Nal data. After completing laboratory testing for functionality arLc stability, initial performance assessment was petf:)rmed using side by side imaging of clinical subject:; at R a m b a m Medical Center, Haifa. Patients undergoing bone studies on an SP6HR, a standard high resolution (95 pmt's) NaI g a m m a camera equipped with a HR collimator, were imaged with the sarrLc view and same acquisition times ( - 5 minutes) on the C Z T system a few minuets later. Scint~inammography performance was tested at the Mayo Clinic, MN, using a water p h a n t o m and a set of small tillable balls of different sizes, immersed -o different levels, and in varying depths of wat,r:. The parameter space is shown in Table 1. Acqui:dtions using the p h a n t o m were made side by side with the CZT camera and a Helix g a m m a
camera to compare lesion detectability over the entire parameter space. Preliminary investigations of clinical scintimammography were also carried out at the Mayo Clinic. A direct comparison image on the CZT camera for a patient undergoing scintimammography was acquired. In addition a clinical trail involving 100 patients referred to CZT-scintimammography from X-ray m a m m o g r a p h y was commenced. Referral included patients with positive m a m m o g r a p h y (grade 4-5) and detected lesions of size 5-20 m m . Images were acquired using X-ray m a m m o g r a p h i c imaging positions. For each patient, cranio-caudal and/ or medio-lateral oblique views of 10 minutes each were acquired of each breast, with light (pain free) compression. 3. RESULTS
Figure 1 shows bone extremity images acquired with the C Z T camera and the SP6HR under the same conditions. The CZT images permit superior visualization of various features, for example the articular surfaces of the knee, the definition of the patella, and the metatarsal bones of the foot. The CZT images begin to have the character of X-rays images. The breast p h a n t o m lesion detectability p h a n t o m was imaged with the CZT and the Helix systems. The systems had similar sensitivities, therefore the acquisitions times and counts were also close. The main differences visible in the images are attributable
TAB. 1. Phantom Variables.
Phantom
Scintimammography
Ball diameter Water depth Depth of balls lAbj : [awater] Acquisition time Collimator
Size of tumor Tissue thickness Tumor location Tumor drug uptake Count density Collimator
Range
Units
5-7
cm cm
1-7 3:1-12:1 0.9, 2.2 GP / HR
Mcnts/ 400cm2
FIG. 1. Clinical comparison of CZT and SP6HR. 57
Ira M. Blevis et
alii:
CZT Gamma Camera for Scintimammography
to differences in the system resolution derived from the pixelation and collimator combination. Figure 2 shows a comparison of images for three values of the depths of the balls. In all cases the CZT shows higher clarity. This clarity is highlighted in the third comparison pair where the 4.2 m m ball and the 4.9 m m ball disappear from the NaI image. In randomized observer trails based on these images a systematic improvement in the n u m b e r of balls and thereby the (small) limit of detectability was shown by the CZT camera, particularly for balls 5-7 m m in diameter. The first clinical CZT-scintimammography comparison is shown in Figure 3. In this case the patient was being checked for reemergence of the primary t u m o r a year after surgical removal. The US and MRI were inconclusive due to the previous tissue damage; the NaI alone would have failed to detect the uptake of 99mTc;however, the CZT image taught the correct interpretation of the NaI image. The clinical trial of CZT-scintimammography has examined 80 of the 100 patients intended. At the time of writing the results were compiled for 40 of the studies. The 40 patients included 36 tumors. Twentysix patients had a total of 36 lesions confirmed at surgery Of these 36 lesions, 33 were detected by the C Z T system (overall sensitivity 92%). There were 22 lesions < 10 m m in diameter (Avg. 8.1 + 1.9 ram), and 19 were detected on C Z T imaging (sensitivity 86%). Two patients had false negative studies. In these pa-
FIG. 2. Breast Lesion Detectability Phantom Images. 58
tients, three lesions that were located deep within the breast were missed primarily due to (future avoidable) technical errors in positioning. M1 14 lesions > 1 cm in diameter seen on the C Z T system were confirmed at surgery. In three patients, lesions were detected by the CZT system in the contralateral breast to the breast containing the suspicious lesion seen on mammography. These lesions were seen on MRI and all were confirmed as true positive cases at surgery A total of 14 patients had negative findings at biopsy or surgery. Of these 9 were true negative and 5 were false positive on the CZT. False positive studies were benign fibroadenoma (3 patients), inflammatory fatty necrosis (1 patient), and benign breast parenchyma (1 patient). Figure 4 shows an example of the image quality realized in these studies. 4. CONCLUSIONS
A promising solution to the problems of dense breast tissue and possibly false positive studies in Xray m a m m o g r a p h y has been proposed in the form of CZT-Scintimammography. Preliminary investigations have been completed. The image quality for close objects is superior to that of NaI cameras in a variety of tests. The sensitivity and specificity is high from these early results. Small lesions in breast cancer < 1 cm should be reliably detected by the technique. Complete results will be available with the finish of the first clinical trial.
~Physica Medica~ - Vol. XXI, Supplement 1, 2006
Rt. b r e a s t - I D C 5 mm diameter
Lateral
Medial
Fro. 4. CZT-scintimammography: Clinical Trial. FIG, 3. CZT-scintimammography Comparison. R E F E R E N CES
[1] [2]
Gra owJ. Breast Cancer 2004, Progress and Promise on the :21inical Front, In this Issue. Kha Ikhali I, Villanueva-Meyer J, Edell SL, et alii Diagnostic ~ :curacy of 99mTc-sestaMIBI breast imaging: multiceiu :r trial results. J Nucl Med, 2000: 41; 1973-1979.
[3] [4]
Blevis I, Reznik A, Pansky A, Liss J, Wainer N. Small Field of View CZT Gamma Camera, IEEE Medical Imaging Conference, 2002. Mueller B, O'Connor M K, Blevis I, Rhodes D J, Smith R, Collins D A, PhillipsJ. Evaluation of a Cadmium Zinc Telluride Detector for Scintimammography, J Nuc Med, 2003: 44; 602-609
59