Motion-Compensated 4D Processing of Gated SPECT Perfusion Studies

† Research supported by NIH/NHLBI Grant HL65425. Abstract—In this paper we present two 4D motion-compensated reconstruction algorithms for gated cardiac SPECT. The algorithms are based on a deformable content-adaptive mesh model, a compact image representation that is well suited for motion tracking of non-rigid objects like the heart. In this paper we introduce a reconstruction method using interiteration smoothing, in which motion estimation, reconstruction, and smoothing are all performed in the mesh domain. We also describe a pixel-based post-reconstruction temporal smoothing approach. In these methods, we modify our previous approaches to account for apparent brightening due to wall thickening (an effect of finite resolution). Our experiments show that the new algorithms produce image sequences that closely match the expected intensity variations at the spatial resolutions of interest. In gated SPECT, these variations are a useful and important clinical indicator of wall thickening, so it is beneficial to preserve them. In our experiments, the proposed methods outperform those that do not take into account the brightening effect.