UNLABELLED: Infusional brachytherapy for treatment of neoplasms, with colloidal 32P has been used to treat various tumors in the pancreas, liver, brain, lung, and head and neck. In performing such treatments, anatomical verification of the location of the administered 32P from the image obtained by Bremsstrahlung SPECT alone is not possible due to the lack of internal landmarks, since the radionuclide is distributed only in the tumor and does not usually accumulate in the normal organs. The purpose of this study was to provide a practical three-dimensional approach for image fusion between Bremsstrahlung SPECT and CT. METHODS: The tumors in four cancer patients were injected directly with 32P under CT guidance. A Bremsstrahlung SPECT study using 99mTc backscatter sources to obtain the body contour was then performed. SPECT images were used to generate the skin contours using a threshold detection method. A three-dimensional surface was generated from these contours using a tiling program and fused with a corresponding CT surface generated from a CT scan in the same patient through an iterative surface-fitting algorithm. The three-dimensional surface of the region of high-activity, corresponding to the infused tumor, was then generated using the Bremsstrahlung SPECT data by mapping the iso-count surfaces through a computer program. The three-dimensional image of the organ then was fused with the registered CT-SPECT datasets. RESULTS: The accuracy of fit measured as the mean distance between the SPECT and CT surfaces was in the range of 3-4 mm. CONCLUSION: The anatomical co-registration of Bremsstrahlung SPECT with CT images using the outer surface-fitting algorithm is a reliable tool. This correlation permits direct anatomic confirmation of the region of the 32P activity distribution with the anatomic site selected for injection.
UNLABELLED: Infusional brachytherapy for treatment of neoplasms, with colloidal 32P has been used to treat various tumors in the pancreas, liver, brain, lung, and head and neck. In performing such treatments, anatomical verification of the location of the administered 32P from the image obtained by Bremsstrahlung SPECT alone is not possible due to the lack of internal landmarks, since the radionuclide is distributed only in the tumor and does not usually accumulate in the normal organs. The purpose of this study was to provide a practical three-dimensional approach for image fusion between Bremsstrahlung SPECT and CT. METHODS: The tumors in four cancerpatients were injected directly with 32P under CT guidance. A Bremsstrahlung SPECT study using 99mTc backscatter sources to obtain the body contour was then performed. SPECT images were used to generate the skin contours using a threshold detection method. A three-dimensional surface was generated from these contours using a tiling program and fused with a corresponding CT surface generated from a CT scan in the same patient through an iterative surface-fitting algorithm. The three-dimensional surface of the region of high-activity, corresponding to the infused tumor, was then generated using the Bremsstrahlung SPECT data by mapping the iso-count surfaces through a computer program. The three-dimensional image of the organ then was fused with the registered CT-SPECT datasets. RESULTS: The accuracy of fit measured as the mean distance between the SPECT and CT surfaces was in the range of 3-4 mm. CONCLUSION: The anatomical co-registration of Bremsstrahlung SPECT with CT images using the outer surface-fitting algorithm is a reliable tool. This correlation permits direct anatomic confirmation of the region of the 32P activity distribution with the anatomic site selected for injection.