Reproduce “Diffeomorphic Mesh Deformation via Efficient Optimal Transport for Cortical Surface Reconstruction” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Tung Le, Khai Nguyen, Shanlin Sun, Kun Han, Nhat Ho, Xiaohui Xie

Researchers from the University of California, Irvine and The University of Texas at Austin introduce DDOT, a diffeomorphic mesh deformation framework for cortical surface reconstruction that utilizes Sliced Wasserstein Distance (SWD) on oriented varifolds as a novel loss function. This approach dramatically reduces self-intersections by a factor of 100 compared to previous state-of-the-art methods, while also achieving superior geometric accuracy and computational efficiency.

Key insights

  • Replacing Chamfer Distance with Sliced Wasserstein Distance (SWD) significantly improves the robustness and efficiency of mesh deformation optimization, especially when comparing complex geometric structures.
  • Representing meshes as oriented varifolds and then as probability measures provides a richer and more effective input for optimal transport-based comparison than traditional point cloud representations.
  • The combination of diffeomorphic flows (via Neural ODEs) with a statistically sound and efficient metric like SWD virtually eliminates self-intersections while maintaining high geometric accuracy.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.