R$^{2}$Seg: Training-Free OOD Medical Tumor Segmentation via Anatomical Reasoning and Statistical Rejection

TL;DR

R$^{2}$Seg is a training-free, two-stage framework that enhances out-of-distribution tumor segmentation in medical images by combining anatomical reasoning with statistical rejection, significantly reducing false positives.

Contribution

It introduces a novel training-free approach that integrates anatomical reasoning and statistical testing for robust OOD tumor segmentation without parameter updates.

Findings

Significantly improves Dice scores on multi-center benchmarks

Reduces false positives and increases specificity

Compatible with zero-update test-time augmentation

Abstract

Foundation models for medical image segmentation struggle under out-of-distribution (OOD) shifts, often producing fragmented false positives on OOD tumors. We introduce R$^{2}$Seg, a training-free framework for robust OOD tumor segmentation that operates via a two-stage Reason-and-Reject process. First, the Reason step employs an LLM-guided anatomical reasoning planner to localize organ anchors and generate multi-scale ROIs. Second, the Reject step applies two-sample statistical testing to candidates generated by a frozen foundation model (BiomedParse) within these ROIs. This statistical rejection filter retains only candidates significantly different from normal tissue, effectively suppressing false positives. Our framework requires no parameter updates, making it compatible with zero-update test-time augmentation and avoiding catastrophic forgetting. On multi-center and multi-modal tumor segmentation benchmarks, R$^{2}$Seg substantially improves Dice, specificity, and sensitivity over strong baselines and the original foundation models. Code are available at https://github.com/Eurekashen/R2Seg.