Adversarial Attacks on Medical Segmentation Model via Transformation of Feature Statistics

Deep learning-based segmentation models have made a profound impact on medical procedures, with U-Net based computed tomography (CT) segmentation models exhibiting remarkable performance. Yet, even with these advances, these models are found to be vulnerable to adversarial attacks, a problem that eq...

Full description

Saved in:
Bibliographic Details
Published in:Applied sciences Vol. 14; no. 6; p. 2576
Main Authors: Lee, Woonghee, Ju, Mingeon, Sim, Yura, Jung, Young Kul, Kim, Tae Hyung, Kim, Younghoon
Format: Journal Article
Language:English
Published: Basel MDPI AG 01.03.2024
Subjects:
adversarial attacks
computed tomography (CT) segmentation
CT imaging
data augmentation
Deep learning
deep learning-based segmentation
Medical research
Medical screening
Methods
realistic adversarial samples
Standard deviation
Statistics
Tomography
Germany
ISSN:2076-3417, 2076-3417
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Deep learning-based segmentation models have made a profound impact on medical procedures, with U-Net based computed tomography (CT) segmentation models exhibiting remarkable performance. Yet, even with these advances, these models are found to be vulnerable to adversarial attacks, a problem that equally affects automatic CT segmentation models. Conventional adversarial attacks typically rely on adding noise or perturbations, leading to a compromise between the success rate of the attack and its perceptibility. In this study, we challenge this paradigm and introduce a novel generation of adversarial attacks aimed at deceiving both the target segmentation model and medical practitioners. Our approach aims to deceive a target model by altering the texture statistics of an organ while retaining its shape. We employ a real-time style transfer method, known as the texture reformer, which uses adaptive instance normalization (AdaIN) to change the statistics of an image’s feature.To induce transformation, we modify the AdaIN, which typically aligns the source and target image statistics. Through rigorous experiments, we demonstrate the effectiveness of our approach. Our adversarial samples successfully pass as realistic in blind tests conducted with physicians, surpassing the effectiveness of contemporary techniques. This innovative methodology not only offers a robust tool for benchmarking and validating automated CT segmentation systems but also serves as a potent mechanism for data augmentation, thereby enhancing model generalization. This dual capability significantly bolsters advancements in the field of deep learning-based medical and healthcare segmentation models.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2076-3417
2076-3417
DOI:10.3390/app14062576