Reproduce “TRACE: Topology-Aware Reconnection via Prediction-Adaptive Connectivity Enforcement for Thin-Structure Segmentation” 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

Junho Moon, Ikbeom Jang

TRACE trains thin-structure segmentation with a topology-aware skeleton carrier that preserves annotated connectivity and locates prediction-dependent local reconnection bottlenecks. In a controlled nnU-Net comparison on DRIVE, it improves clDice (82.65 versus 81.89) and lowers Betti-number error (79.30 versus 97.00) against Skeleton Recall Loss, while Dice remains similar (82.38 versus 82.18). Results cover six benchmarks; annotations’ topology, including errors, is retained.

Summary by alphaXiv.

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