Multi-contrast quantitative mapping with an unsupervised reconstruction method based on implicit neural representation

Motivation: Multi-contrast quantitative MRI usually requires multiple scans, leading to long acquisition time and potential inter-scan misalignment. Goal(s): To achieve the multi-contrast quantitative MRI acquisition in a single scan and improve the accuracy of the quantitative mapping. Approach: We developed a multi-contrast quantitative mapping sequence to simultaneously obtain T1, T2, T2* maps and subvoxel QSM. Reconstruction was conducted to directly estimate the underlying quantitative maps from the highly undersampled high-dimensional k-space data. The proposed framework was validated on the simulation, phantom and healthy volunteers. Results: The results demonstrated that our proposed method exhibited a high correlation with references on the quantitative maps. Impact: The proposed acquisition and reconstruction framework can simultaneously provide multi-contrast quantitative maps of the whole brain within a 5.8-minute scan. This new technique is clinically promising for tissue characterization and pathological research in neurosciences.