On the use of prior information by maximum likelihood reconstructions in emission tomography

We examine the usefulness of prior Information in emission tomography. Maximum likelihood reconstruction algorithms are used in the study. We attempt to answer the question whether prior information on the contents of some areas of an image improves the efficacy of the reconstruction in other areas of the image. We use mathematical phantoms based on the human of brain. Our results show that the prior information that there is no activity outside the brain does not improve the efficacy of reconstructions, but having prior information about the whole brain except one neuroanatomical structure (the region of interest, ROI for short) significantly improves the capability of estimating the activity in the ROI.