A self-navigated framework for motion detection and spoke-reordering musculoskeletal dynamic reconstruction in 3D radial MRI.

Motion remains a major challenge and active research focus in MRI, as both involuntary (e.g., head movement) and voluntary (e.g., joint motion) motion can degrade image quality or, alternatively, provide opportunities for dynamic assessment. Existing motion sensing approaches - such as external tracking or navigator sequences - often require additional hardware, increase SAR, or necessitate sequence modifications, thereby limiting clinical flexibility. In this work, we propose a self-navigated motion sensing framework for 3D radial MRI. Motion surrogates are first extracted directly from multi-coil k-space data. A sliding-window summation is then applied to enhance robustness, followed by a second principal component (PC2)-based strategy to derive a 1D motion-sensitive signal that reflects subject position over time. Beyond conventional head motion correction, we demonstrate the utility of the proposed framework in improving dynamic 4D MRI of the ankle and knee under both stepwise and continuous motion protocols. By reordering spokes based on motion-derived position rather than acquisition time, the method enables motion-resolved reconstructions with improved anatomical clarity. Overall, this approach facilitates efficient motion detection and supports broader adoption of motion-aware dynamic MRI. The source code of this framework is available at https://github.com/EricLin1993/SpEMSK.