Regularization based Modified SART Iterative Method for CT Image Reconstruction

Simultaneous algebraic reconstruction Technique (SART) based iterative method play a significant role in the quality of images reconstructed by Computed Tomography (CT). The basic limitations associated with this method include slow convergence, poor visual quality and ill-posedness. To address these drawbacks, SART iterative method is modified in two steps. The first step removes the issue of slow convergence and poor visual quality by incorporating one-step-late (OSL) technique in SART referred as modified SART (SART+OSL). The result of first step undergoes in Anisotropic Diffusion (AD) as regularization prior to deal with illposedness issue. To evaluate the proposed method, both qualitative and quantitative studies were conducted and results were compared with existing methods using two different simulated test phantoms. Experimental results show that the proposed model yields significant gain in terms of visual reconstructed image quality and noise suppression.