Robust convolutional neural networks against adversarial attacks on medical images

2022/07/22 by Xiaoshuang Shi, Yifan Peng, Qingyu Chen +8 · 47 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · #Adversarial Robustness in Machine Learning #Adversarial system #Anomaly Detection Techniques and Applications #Artificial intelligence #Bacillus and Francisella bacterial research #Computer science #Computer security #Convolutional neural network #Deep learning #Discriminative model #Embedding #Machine learning #Modalities #Pattern recognition (psychology) #Robustness (evolution) #Vulnerability (computing)

paper · pdf · doi:10.1016/j.patcog.2022.108923

published in Pattern Recognition 132, 108923 (Elsevier BV)

openalex publication_date 2022/07/22 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/02

Abstract

Convolutional neural networks (CNNs) have been widely applied to medical images. However, medical images are vulnerable to adversarial attacks by perturbations that are undetectable to human experts. This poses significant security risks and challenges to CNN-based applications in clinic practice. In this work, we quantify the scale of adversarial perturbation imperceptible to clinical practitioners and investigate the cause of the vulnerability in CNNs. Specifically, we discover that noise (i.e., irrelevant or corrupted discriminative information) in medical images might be a key contributor to performance deterioration of CNNs against adversarial perturbations, as noisy features are learned unconsciously by CNNs in feature representations and magnified by adversarial perturbations. In response, we propose a novel defense method by embedding sparsity denoising operators in CNNs for improved robustness. Tested with various state-of-the-art attacking methods on two distinct medical image modalities, we demonstrate that the proposed method can successfully defend against those unnoticeable adversarial attacks by retaining as much as over 90% of its original performance. We believe our findings are critical for improving and deploying CNN-based medical applications in real-world scenarios.

Citations

• Adversarial attacks on medical machine learning

Cited by

• Comparing Detection Schemes for Adversarial Images against Deep Learning Models for Cancer Imaging. [europepmc]
• Security and Privacy in Machine Learning for Health Systems: Strategies and Challenges. [europepmc]
• Improved Generalizability in Medical Computer Vision: Hyperbolic Deep Learning in Multi-Modality Neuroimaging. [europepmc]
• FungID: Innovative Fungi Identification Method with Chromogenic Profiling of Colony Color Patterns. [europepmc]
• A multi-layered defense against adversarial attacks in brain tumor classification using ensemble adversarial training and feature squeezing. [europepmc]

Related

• Stabilized Medical Image Attacks
• Securing the Diagnosis of Medical Imaging: An In-depth Analysis of AI-Resistant Attacks
• Impact of White-Box Adversarial Attacks on Convolutional Neural Networks
• A Hierarchical Feature Constraint to Camouflage Medical Adversarial Attacks
• Adversarially Robust Medical Classification via Attentive Convolutional Neural Networks