Anatomically-aware conformal prediction for medical image segmentation with random walks

2026-01-26

AI Summary

Standard conformal prediction in medical image segmentation often yields fragmented and spatially inconsistent prediction sets due to its neglect of anatomical context, limiting clinical utility. This work proposes Random Walk Conformal Prediction (RW-CP), a model-agnostic framework that constructs a k-nearest neighbor graph on the outputs of any segmentation model using features from a pretrained vision foundation model, then propagates uncertainty via random walks to regularize nonconformity scores. By integrating random walk dynamics into conformal prediction for the first time, RW-CP substantially enhances the spatial coherence and anatomical plausibility of prediction sets while exhibiting greater robustness to the calibration parameter λ. Evaluated on multiple multimodal public datasets, RW-CP achieves up to a 35.4% improvement in segmentation quality over baseline methods while strictly maintaining marginal coverage at α = 0.1.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

The reliable deployment of deep learning in medical imaging requires uncertainty quantification that provides rigorous error guarantees while remaining anatomically meaningful. Conformal prediction (CP) is a powerful distribution-free framework for constructing statistically valid prediction intervals. However, standard applications in segmentation often ignore anatomical context, resulting in fragmented, spatially incoherent, and over-segmented prediction sets that limit clinical utility. To bridge this gap, this paper proposes Random-Walk Conformal Prediction (RW-CP), a model-agnostic framework which can be added on top of any segmentation method. RW-CP enforces spatial coherence to generate anatomically valid sets. Our method constructs a k-nearest neighbour graph from pre-trained vision foundation model features and applies a random walk to diffuse uncertainty. The random walk diffusion regularizes the non-conformity scores, making the prediction sets less sensitive to the conformal calibration parameter $\lambda$, ensuring more stable and continuous anatomical boundaries. RW-CP maintains rigorous marginal coverage while significantly improving segmentation quality. Evaluations on multi-modal public datasets show improvements of up to $35.4\%$ compared to standard CP baselines, given an allowable error rate of $\alpha=0.1$.

Problem

• conformal prediction
• medical image segmentation
• anatomical coherence
• spatial incoherence
• uncertainty quantification

Innovation

• Conformal Prediction
• Random Walk
• Medical Image Segmentation
• Anatomical Coherence
• Uncertainty Quantification