MRI Image reconstruction using deep learning

Abstract

MRI is a non-invasive imaging modality that provides excellent soft tissue contrast as compared with other imaging techniques e.g., X-ray and CT, etc. MRI comes with the drawback of long scan time due to the slow data acquisition process. Data under-sampling is performed to accelerate the scan time which leads to artifacts. This thesis presents a deep learning-based MR image reconstruction from 1D-Cartesian variable density under-sampled MR image data. The proposed method significantly outperforms all the evaluated approaches, based on a thorough comparison of the proposed methodology with other approaches.