Adversarial Attack Vulnerability of Medical Image Analysis Systems: Unexplored Factors

Abstract

Adversarial attacks are considered a potentially serious security threat for machine learning systems. Medical image analysis (MedIA) systems have recently been argued to be vulnerable to adversarial attacks due to strong financial incentives and the associated technological infrastructure. In this paper, we study previously unexplored factors affecting adversarial attack vulnerability of deep learning MedIA systems in three medical domains: ophthalmology, radiology, and pathology. We focus on adversarial black-box settings, in which the attacker does not have full access to the target model and usually uses another model, commonly referred to as surrogate model, to craft adversarial examples. We consider this to be the most realistic scenario for MedIA systems. Firstly, we study the effect of weight initialization (ImageNet vs. random) on the transferability of adversarial attacks from the surrogate model to the target model. Secondly, we study the influence of differences in development data between target and surrogate models. We further study the interaction of weight initialization and data differences with differences in model architecture. All experiments were done with a perturbation degree tuned to ensure maximal transferability at minimal visual perceptibility of the attacks. Our experiments show that pre-training may dramatically increase the transferability of adversarial examples, even when the target and surrogate's architectures are different: the larger the performance gain using pre-training, the larger the transferability. Differences in the development data between target and surrogate models considerably decrease the performance of the attack; this decrease is further amplified by difference in the model architecture. We believe these factors should be considered when developing security-critical MedIA systems planned to be deployed in clinical practice.

First three authors contributed equally

References in corpus (15)

• Explaining and Harnessing Adversarial Examples
• CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning
• Theoretically Principled Trade-off between Robustness and Accuracy
• Delving into Transferable Adversarial Examples and Black-box Attacks
• Automated Gleason Grading of Prostate Biopsies using Deep Learning
• On Evaluating Adversarial Robustness
• Understanding Adversarial Attacks on Deep Learning Based Medical Image Analysis Systems
• The Space of Transferable Adversarial Examples
• Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
• Computer aided detection of tuberculosis on chest radiographs: An evaluation of the CAD4TB v6 system
• Evaluation of a deep learning system for the joint automated detection of diabetic retinopathy and age-related macular degeneration
• Dropout Inference in Bayesian Neural Networks with Alpha-divergences
• Pseudo-healthy synthesis with pathology disentanglement and adversarial learning
• Deep learning assessment of breast terminal duct lobular unit involution: towards automated prediction of breast cancer risk
• Is AmI (Attacks Meet Interpretability) Robust to Adversarial Examples?

Cited by in corpus (10)

• MedViT: A Robust Vision Transformer for Generalized Medical Image Classification
• How Deep Learning Sees the World: A Survey on Adversarial Attacks & Defenses
• Data synthesis and adversarial networks: A review and meta-analysis in cancer imaging
• Survey on Adversarial Attack and Defense for Medical Image Analysis: Methods and Challenges
• When and How to Fool Explainable Models (and Humans) with Adversarial Examples
• Adversarial Attack Driven Data Augmentation for Accurate And Robust Medical Image Segmentation
• Improving Robustness and Reliability in Medical Image Classification with Latent-Guided Diffusion and Nested-Ensembles
• CoRPA: Adversarial Image Generation for Chest X-rays Using Concept Vector Perturbations and Generative Models
• Revisiting Hidden Representations in Transfer Learning for Medical Imaging
• Universal and Transferable Attacks on Pathology Foundation Models