Reproduce “OcTreeMamba: Topology-Aware Mamba for 3D Medical Segmentation via Octree Sparsification” 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
Kaixiang Mai, Jiaqi Yu, Yuzhu Ji, Yiqun Zhang, Zhijie Wang, Haijun Zhang, Yiu-ming Cheung
OcTreeMamba segments 3D medical images by selecting active regions, arranging them in a spatial hierarchy, and passing context through bidirectional Tree-SSM propagation. It fuses sparse features into a dense U-Net pathway. On ImageCAS, the bottleneck-only model reports Dice 82.31, clDice 87.67, and HD95 12.88, versus nnU-Net’s 81.54, 86.60, and 16.03; this is benchmark evidence, not a universal advantage.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.