Imbalanced Medical Image Segmentation with Pixel-dependent Noisy Labels

TL;DR

This paper introduces CLCS, a novel framework for medical image segmentation that effectively handles pixel-dependent noisy labels and class imbalance through collaborative learning, dynamic thresholding, and noise-aware loss functions.

Contribution

The proposed CLCS framework uniquely addresses pixel-dependent noisy labels and class imbalance with a collaborative learning approach and adaptive thresholding, improving segmentation accuracy.

Findings

Outperforms existing methods on medical image segmentation tasks.

Effectively utilizes noisy labels through noise balance loss.

Improves robustness to label noise and class imbalance.

Abstract

Accurate medical image segmentation is often hindered by noisy labels in training data, due to the challenges of annotating medical images. Prior research works addressing noisy labels tend to make class-dependent assumptions, overlooking the pixel-dependent nature of most noisy labels. Furthermore, existing methods typically apply fixed thresholds to filter out noisy labels, risking the removal of minority classes and consequently degrading segmentation performance. To bridge these gaps, our proposed framework, Collaborative Learning with Curriculum Selection (CLCS), addresses pixel-dependent noisy labels with class imbalance. CLCS advances the existing works by i) treating noisy labels as pixel-dependent and addressing them through a collaborative learning framework, and ii) employing a curriculum dynamic thresholding approach adapting to model learning progress to select clean data samples to mitigate the class imbalance issue, and iii) applying a noise balance loss to noisy data samples to improve data utilization instead of discarding them outright. Specifically, our CLCS contains two modules: Curriculum Noisy Label Sample Selection (CNS) and Noise Balance Loss (NBL). In the CNS module, we designed a two-branch network with discrepancy loss for collaborative learning so that different feature representations of the same instance could be extracted from distinct views and used to vote the class probabilities of pixels. Besides, a curriculum dynamic threshold is adopted to select clean-label samples through probability voting. In the NBL module, instead of directly dropping the suspiciously noisy labels, we further adopt a robust loss to leverage such instances to boost the performance.