A New Algorithm for Image Reconstruction for Positron Emission Tomography

In Positron Emission Tomography, penalized iterative algorithms like MAP often results in over smooth reconstructions due to over penalizing nature of the assumed interacting potential. These algorithms fail to determine the density class of the estimate and hence penalize the pixels irrespective of the density class. In this work, a fuzzy logic based approach is proposed to model the prior which captures the nature of pixel-pixel interaction. Quantitative analysis shows that the proposed fuzzy rule based reconstruction algorithm is capable of producing comparable estimates when compared to MAP and MRP algorithms. The reconstructed images are sharper in nature and the presence of spatially confined minute features are well reconstructed due to the local nature of the penalizing fuzzy potential.