RT Journal Article SR Electronic T1 Optimized projection ordering schemes for Compton camera reconstruction using OSEM JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 1479 OP 1479 VO 50 IS supplement 2 A1 Soo Mee Kim A1 Jae Sung Lee A1 Chun Sik Lee A1 Chan Hyeong Kim A1 Myung Chul Lee A1 Dong Soo Lee A1 Soo-Jin Lee YR 2009 UL http://jnm.snmjournals.org/content/50/supplement_2/1479.abstract AB 1479 Objectives Although the OSEM algorithm does not converge to the true ML solution, it is known to provide a good solution if the projections that constitute each subset are reasonably balanced. Here we investigate three different projection ordering schemes of OSEM for Compton cameras, which can efficiently produce subsets in a balanced way. Methods The Compton scattered data can be subdivided into subsets by a combination of scatting angles (SA) and detected positions (DP). To construct subsets which are balanced so that voxel activity contributes equally to any subset, the data were first arranged by three ordering schemes; the random ordering scheme (ROS), the multilevel ordering scheme (MOS), and the weighted-distance ordering scheme (WOS). The arranged data were then split into N subsets. To compare three ordering schemes, we calculated the coefficient of variation (CV) of the angular and positional differences and the percentage errors (PE). Results All ordering schemes showed the order-of-magnitude acceleration over the standard EM, and their computation times were about the same. The MOS and WOS provided the maximum mean-difference in case of SA- and DP-based arrangement, respectively. In contrast, the WOS exhibited minimum CVs for both SA- and DP-based arrangements. Conclusions The WOS that yielded the best results with the lowest PEs by providing larger and more uniform angular and positional differences in the SA- and DP-arrangements would be the optimal choice for Compton camera reconstruction using OSEM.