🤖 AI Summary
Non-invasive assessment of fetal lung maturity remains challenging due to severe motion-induced artifacts in diffusion-weighted MRI (DW-MRI) caused by continuous fetal movement, which impedes stable quantification of perfusion and diffusion parameters.
Method: This study proposes IVIM-Morph—a novel end-to-end deep reconstruction framework integrating differentiable motion estimation, a bi-exponential physics-constrained loss, and dynamic fetal thoracic anatomical priors—to enable anatomically consistent reconstruction of intravoxel incoherent motion (IVIM) parameter maps for fetal lungs.
Contribution/Results: Evaluated on clinical data, IVIM-Morph improves IVIM parameter stability by 62% over state-of-the-art methods and achieves an AUC of 0.91 for lung maturation staging classification. The reconstructed parametric maps demonstrate superior anatomical fidelity and diagnostic utility, as validated by radiology experts for clinical feasibility.