Official preview for the MICCAI 2026 paper
Sulcus-Aware Hippocampal Surface Modeling with Training-time Sulcus-guided Learning
in the
Sulcus-Aware-Hippo-Surface
repository.
The proposed method reconstructs topology-consistent hippocampal surfaces by deforming a sulcus-preserving template to individual subjects. It uses training-time sulcus-guided learning to improve sulcal fidelity while maintaining stable surface topology.
Hippocampal surface analysis can capture localized structural variation that is not fully represented by volumetric summaries. However, conventional reconstruction pipelines may suppress the hippocampal sulcus or introduce unstable topology. This method transfers a sulcus-preserving template surface to subject anatomy through a learned smooth deformation field, enabling correspondence-aware surface modeling with improved sulcal preservation.
Code and pretrained resources are coming soon.