Hebbian Learning Based Image Reconstruction for Positron Emission Tomography

Maximum a-posteriori (MAP) algorithms eliminates noisy artifacts by utilizing available prior information in the reconstruction process. The MAP based algorithms fail to determine the density class in the reconstructed image and hence penalize the pixels irrespective of the density class and irrespective of interaction between the nearest neighbors. In this paper, Hebbian neural learning scheme is proposed to model the nature of inter-pixel interaction in order to reconstruct artifact-free edge-preserving reconstruction. It is assumed that local correlation plays a significant role in the image reconstruction process and proper modeling of correlation weight (which defines the strength of inter-pixel interaction) is essential for generating artifact free reconstruction. Quantitative analysis shows that the proposed scheme based reconstruction algorithm is capable of producing better reconstructed images. The reconstructed images are sharper with small features being better resolved