A proposal and evaluation of spatio-temporal reconstruction method based on DRAMA

Parametric images are very useful to know spatial and temporal changes and differences among regions of tracer. In the general approaches to estimate the parametric images, the tissue Time Activity Curves (tTACs) are obtained from each pixels of a reconstructed dynamic PET image at first. However the reconstruction time of the tTACs becomes huge, this tendency becomes remarkable increasingly in 3D DOI-PET, and the accuracy of the parametric images strongly depends on the accuracy of tTACs. The spatio-temporal PET reconstruction methods have been studied for now but previous methods cannot achieve enough accuracy yet and they also require much computational time. In this paper, we propose a new spatiotemporal reconstruction method based on Dynamic Row-Action Maximum-likelihood Algorithm (DRAMA). Our method requires only one reconstructing process and dramatically reduces the computational demand to estimate the tTACs. Algorithm performance is evaluated using digital phantoms and the results show that the proposed method produces parametric images with better quality than that by a conventional method, with less computation than that of conventional one.