We have just published a new paper in APSIPA Transactions on Signal and Information Processing on “Self-Supervised Motion-Corrected Image Reconstruction Network for 4D Magnetic Resonance Imaging of the Body Trunk”. This paper is an extension of our award winning APSIPA 2020 conference paper.

The paper presents a deep-learning based motion-corrected 4D (3D spatial + time) image reconstruction for magnetic resonance imaging (MRI). The approach combines a non-rigid registration network and a 4D reconstruction network. The non-rigid registration network is based on our local all-pass filter framework for image registration which was extended to MRI data here.

The paper is available here: Self-Supervised Motion-Corrected Image Reconstruction Network for 4D Magnetic Resonance Imaging of the Body Trunk.

Proposed motion-compensated 4D reconstruction deep learning network