A subject-specific reversible folding model reveals geometry-driven white-matter organization

Abstract

Tractography currently relies almost exclusively on diffusion data and seed maps, which alone remain insufficient for anatomically accurate fiber reconstruction. During brain development, white-matter fibers elongate alongside cortical folding, suggesting a close mechanistic link between cortical geometry and fiber organization. To investigate this link, we introduce a subject-specific cortical folding simulation framework that reconstructs an individuals folding trajectory using only a structural T1-weighted image. The quasi-static, constraint-based model reverses folding to generate an unfolded, fetal-like configuration and then refolds to map the resulting volumetric deformation onto fiber organization. Vali-dation with longitudinal fetal MRI shows that simulated folds follow biologically plausible developmental paths. Applied to adult data, deformation of simple radial fibers reproduces diffusion-derived orientation patterns across the white matter and achieves high regional correspondence. The deformation model also generates characteristic short-range U-fibers as well as long-range association and commissural pathways, all without any diffusion in-put or machine learning. This approach provides a new, anatomically grounded source of subject-specific fiber orientation derived solely from cortical geometry, opening avenues for geometry-informed tractography.