Generalizing Abstention for Noise-Robust Learning in Medical Image Segmentation

TL;DR

This paper introduces a universal abstention framework for noise-robust medical image segmentation, improving model reliability by allowing selective ignoring of noisy samples, and demonstrates its effectiveness across multiple loss functions and datasets.

Contribution

The paper proposes a modular abstention framework with an informed regularization and auto-tuning algorithm, enhancing noise robustness in segmentation models across various loss functions.

Findings

Outperforms non-abstaining baselines on CaDIS and DSAD datasets

Effective under high noise levels in medical image segmentation

Versatile integration with multiple loss functions

Abstract

Label noise is a critical problem in medical image segmentation, often arising from the inherent difficulty of manual annotation. Models trained on noisy data are prone to overfitting, which degrades their generalization performance. While a number of methods and strategies have been proposed to mitigate noisy labels in the segmentation domain, this area remains largely under-explored. The abstention mechanism has proven effective in classification tasks by enhancing the capabilities of Cross Entropy, yet its potential in segmentation remains unverified. In this paper, we address this gap by introducing a universal and modular abstention framework capable of enhancing the noise-robustness of a diverse range of loss functions. Our framework improves upon prior work with two key components: an informed regularization term to guide abstention behaviour, and a more flexible power-law-based auto-tuning algorithm for the abstention penalty. We demonstrate the framework's versatility by systematically integrating it with three distinct loss functions to create three novel, noise-robust variants: GAC, SAC, and ADS. Experiments on the CaDIS and DSAD medical datasets show our methods consistently and significantly outperform their non-abstaining baselines, especially under high noise levels. This work establishes that enabling models to selectively ignore corrupted samples is a powerful and generalizable strategy for building more reliable segmentation models. Our code is publicly available at https://github.com/wemous/abstention-for-segmentation.