Dynamic Perturbation-Adaptive Adversarial Training on Medical Image Classification

TL;DR

This paper introduces a dynamic perturbation-adaptive adversarial training method for medical image classification that enhances robustness and generalization of CNNs, outperforming traditional methods on dermatology datasets.

Contribution

The proposed DPAAT method adaptively generates data perturbations and updates criteria dynamically, addressing limitations of fixed perturbation sizes and external transfer dependence in adversarial training.

Findings

Improved robustness and generalization in CNNs for medical imaging.

Significant enhancement in mean average precision and interpretability.

Effective across various CNN architectures.

Abstract

Remarkable successes were made in Medical Image Classification (MIC) recently, mainly due to wide applications of convolutional neural networks (CNNs). However, adversarial examples (AEs) exhibited imperceptible similarity with raw data, raising serious concerns on network robustness. Although adversarial training (AT), in responding to malevolent AEs, was recognized as an effective approach to improve robustness, it was challenging to overcome generalization decline of networks caused by the AT. In this paper, in order to reserve high generalization while improving robustness, we proposed a dynamic perturbation-adaptive adversarial training (DPAAT) method, which placed AT in a dynamic learning environment to generate adaptive data-level perturbations and provided a dynamically updated criterion by loss information collections to handle the disadvantage of fixed perturbation sizes in conventional AT methods and the dependence on external transference. Comprehensive testing on dermatology HAM10000 dataset showed that the DPAAT not only achieved better robustness improvement and generalization preservation but also significantly enhanced mean average precision and interpretability on various CNNs, indicating its great potential as a generic adversarial training method on the MIC.