Bayesian statistics-guided label refurbishment mechanism: Mitigating label noise in medical image classification.

Abstract

Deep neural networks (DNNs) have been widely applied in medical image classification, benefiting from its powerful mapping capability among medical images. However, these existing deep learning-based methods depend on an enormous amount of carefully labeled images. Meanwhile, noise is inevitably introduced in the labeling process, degrading the performance of models. Hence, it is significant to devise robust training strategies to mitigate label noise in the medical image classificationtasks. In this work, we propose a novel Bayesian statistics-guided label refurbishment mechanism (BLRM) for DNNs to prevent overfitting noisy images. BLRMutilizes maximum a posteriori probability in the Bayesian statistics and the exponentially time-weighted technique to selectively correct the labels of noisy images. The training images are purified gradually with the training epochs when BLRMis activated, further improving classificationperformance. Comprehensive experiments on both synthetic noisy images (public OCT & Messidor datasets) and real-world noisy images (ANIMAL-10N) demonstrate that BLRMrefurbishes the noisy labels selectively, curbing the adverse effects of noisy data. Also, the anti-noise BLRMsintegrated with DNNs are effective at different noise ratio and are independent of backbone DNN architectures. In addition, BLRMis superior to state-of-the-art comparative methods of anti-noise. These investigations indicate that the proposed BLRMis well capable of mitigating label noise in medical image classificationtasks.