Towards Fast and Motion-Robust Foot and Knee MRI: Application of SAMER Retrospective Motion Correction with Deep Learning for 2D TSE and 3D SPACE

Motivation: High-resolution imaging is required for detecting subtle pathologies in musculoskeletal MRI, but even slight involuntary patient motion during the lengthy exams can introduce artifacts that may degrade image quality. Goal(s): Evaluate the effectiveness of SAMER retrospective motion detection and correction applied to knee and foot MRI. Approach: SAMER was implemented into PDw and T2w 2D TSE and 3D SPACE sequences, utilizing scout and guidance lines for rapid motion estimation and a deep learning-based reconstruction framework to enhance image quality for highly accelerated acquisitions. Results: Findings indicate significant reductions in motion artifacts and improved accuracy in automated cartilage segmentation in the knee. Impact: SAMER retrospective motion correction was applied to knee and foot imaging and integrated with a deep learning reconstruction to facilitate fast and motion-robust 2D and 3D MSK imaging. This could enhance clinical diagnostics and improve patient outcomes in musculoskeletal MRI.