TV-based DOI De-blurring model for the dual-head flat-panel PET system

摘要

In the dual-head flat panel PET system (DHAPET), the unique detector arrangement leads to severe uncertainty of depth-of-interaction (DOI) and data truncation, which in turn result in poor image quality and data loss in the reconstructed images. To improve the distortions, we proposed a reconstruction strategy under the framework of the fast iterative shrinkage-thresholding algorithm (FISTA) for minimizing total variation and de-blurring the DOI effects. Through the FISTA, the time-consuming procedure of TV-constraint can be accelerated. Additionally, we incorporated the point spread function into the reconstruction procedure of the FISTA to account for the blurring effects of DOI. Our numerical results showed that the proposed method can not only achieve a fast convergence rate to facilitate its applications, but also compensate the DOI effect and artifacts to improve image quality.