QMix: Quality-aware Learning with Mixed Noise for Robust Retinal Disease Diagnosis

TL;DR

This paper introduces QMix, a novel noise learning framework that enhances the robustness of retinal disease diagnosis models by effectively handling mixed noise, including label and data noise, through sample separation and quality-aware semi-supervised training.

Contribution

QMix is the first framework to address mixed noise in medical images by combining sample separation with quality-aware semi-supervised learning, improving robustness and accuracy.

Findings

QMix outperforms existing methods on five retinal datasets.

It significantly improves robustness against mixed noise.

The approach effectively separates high and low quality mislabeled images.

Abstract

Due to the complexity of medical image acquisition and the difficulty of annotation, medical image datasets inevitably contain noise. Noisy data with wrong labels affects the robustness and generalization ability of deep neural networks. Previous noise learning methods mainly considered noise arising from images being mislabeled, i.e. label noise, assuming that all mislabeled images are of high image quality. However, medical images are prone to suffering extreme quality issues, i.e. data noise, where discriminative visual features are missing for disease diagnosis. In this paper, we propose a noise learning framework, termed as QMix, that learns a robust disease diagnosis model under mixed noise. QMix alternates between sample separation and quality-aware semisupervised training in each training epoch. In the sample separation phase, we design a joint uncertainty-loss criterion to effectively separate (1) correctly labeled images; (2) mislabeled images with high quality and (3) mislabeled images with low quality. In the semi-supervised training phase, we train a disease diagnosis model to learn robust feature representation from the separated samples. Specifically, we devise a sample-reweighing loss to mitigate the effect of mislabeled images with low quality during training. Meanwhile, a contrastive enhancement loss is proposed to further distinguish mislabeled images with low quality from correctly labeled images. QMix achieved state-of-the-art disease diagnosis performance on five public retinal image datasets and exhibited substantial improvement on robustness against mixed noise.