Iterative region of interest reconstruction in emission tomography

This paper proposes a new method for region of interest reconstruction in emission tomography assuming projections from the entire field of view (FOV) are available. Unlike some other region of interest evaluation approaches where the system matrix is reformulated simply by summations over columns corresponding to pixels outside the region of interest, our approach reformulates the system matrix using the Bayes probability formula. The simulation study reveals that our method outperforms the traditional approaches judged by reconstruction quality.