Motivation: Synthetic MRI derived from quantitative MRI scans do not always synthetize accurate contrasts for some type of sequences (e.g., FLAIR, post-gadolinium T1w). Goal(s): To achieve accurate synthetic MRI within a short scan to accelerate standard clinical protocols. Approach: We developed an ultra-highly undersampled compressed sensing (CS) scheme that can generate accurate contrast for FLAIR and post-gadolinium T1w sequences, utilizing a patient specific physics-based synthetic MRI as prior in the CS reconstruction. Results: The proposed hybrid CS synthetization of FLAIR and post-gadolinium T1w scans from five neurological patients indicate improved tissue depiction, with a particular emphasis on lesion enhancement and overall anatomical structure. Impact: This improved synthesis approach, using an ultra-highly undersampled hybrid CS protocol, achieves more accurate contrast synthetization and enhances the potential of synthetic MRI for shorter clinical scan times.
Xu et al. (2025) studied this question.