Deep Neural Network Assisted Iterative Reconstruction Method for Low\n Dose CT

Low Dose Computed Tomography suffers from a high amount of noise and/or\nundersampling artefacts in the reconstructed image. In the current article, a\nDeep Learning technique is exploited as a regularization term for the iterative\nreconstruction method SIRT. While SIRT minimizes the error in the sinogram\nspace, the proposed regularization model additionally steers intermediate SIRT\nreconstructions towards the desired output. Extensive evaluations demonstrate\nthe superior outcomes of the proposed method compared to the state of the art\ntechniques. Comparing the forward projection of the reconstructed image with\nthe original signal shows a higher fidelity to the sinogram space for the\ncurrent approach amongst other learning based methods.\n