Motion-compensated implicit neural modeling for 3D multiparametric quantitative MRI

Multiparametric quantitative MRI (MP-qMRI) provides comprehensive 3D tissue characterization in neuroimaging but remains highly susceptible to involuntary head motion. Motion correction in 3D MP-qMRI is particularly challenging, as even subtle head movements corrupt volumetric encoding, disrupt inter-contrast consistency, and bias quantitative parameter estimation. In this work, we propose a motion-corrected MP-qMRI framework that integrates rapid navigator-based motion tracking with motion-compensated implicit neural modeling. A spatiotemporal controlled aliasing (k-t CAIPI) navigator acquisition enables efficient estimation of time-resolved rigid-body motion. The implicit neural formulation embeds these estimates into the signal model to recover motion-corrected quantitative maps. The framework is evaluated through retrospective, simulation, and in vivo experiments, demonstrating substantial reductions in motion-induced artifacts and improved quantitative accuracy across T1, T2, and T2* maps. Evaluation in a motion-prone patient with spinocerebellar ataxia type 3 further highlights the robustness of the approach under clinically challenging conditions. Overall, this work establishes a principled and flexible approach for addressing motion in 3D MP-qMRI, providing a generalizable strategy for motion-resilient quantitative neuroimaging.