Leveraging Acquisition Knowledge to Enhance Robustness of Physics-Informed Motion Correction for T2* Quantification

Motivation: T2*-quantification from GRE-MRI is particularly impacted by subject motion due to its sensitivity to field inhomogeneities. We have previously introduced PHIMO for physics-informed motion correction, which detects and excludes motion-corrupted k-space lines from a learning-based reconstruction. Goal(s): Improving PHIMO's performance for challenging motion patterns and increasing its detection robustness. Approach: We propose to keep the four central k-space points to avoid signal loss when excluding central k-space lines, and exploit the interleaved acquisition scheme by optimizing only one exclusion mask for even and odd slices for a robust brain-wide line detection. Results: The proposed extensions improve PHIMO's robustness and image quality. Impact: Our proposed extensions enhance PHIMO's reconstruction performance and line detection robustness, making T2*-quantification more reliable under various motion conditions. The enhanced PHIMO reaches the performance of a state-of-the-art correction method, while accelerating the acquisition by over 40%, facilitating clinical applicability.