Dynamic imaging on the high resolution research tomograph (HRRT): non-human primate studies

The high resolution research tomography (HRRT) is currently the most complex human brain scanner due to its ability to detect the gamma depth of interaction, its octagonal geometry, and the large number of crystals (119,808) leading to approximately 4.5 /spl times/ 10/sup 9/ possible lines of response (LORs). Reconstruction of dynamic studies on this scanner is particularly challenging due to the dynamic range of both, number of acquired events per frame and acquisition count rates. Some artifacts have been observed with phantom studies: here we evaluate their impact on time activity curves (TACs) and binding potential (BP) values in realistic scanning situations with the ultimate goal of defining an efficient and accurate image reconstruction protocol. Non-human primate studies were used for this purpose. We compared TACs and BPs obtained from images reconstructed with three different reconstruction algorithms, two different axial spanning configurations and detector normalization factors obtained from two different data sets. We also compared BP values obtained from scans of the same animal performed on the Siemens ECAT 963B and the HRRT under identical conditions. The statistical reconstruction methods produced nearly identical results and the impact of emission/normalization count rate mismatch was found to be effectively negligible. Likewise no image degradation due to increased axial spanning was observed. Data obtained from the analytical method were less robust and in general much more sensitive to noise, thus demonstrating a suboptimal performance of this algorithm. The BP values obtained with the HRRT were by approximately 50% higher compared to those obtained in the ECAT as a result of the increased resolution of this tomograph.