Adversarial Attacks on Medical Segmentation Model via Transformation of Feature Statistics
Applied Sciences · 19.03.2024 · 14 (6) · 2576
DOI: 10.3390/app14062576 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 19.03.2024. Verfasst von Woonghee Lee und 5 weiteren Autor:innen. Die Arbeit wurde bisher 8-mal zitiert (Crossref) und nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministry of Trade, Industry and Energy (MOTIE), Korea government (MSIT), Ministry of Education (MOE, Korea).
Abstract (Originalsprache)
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.
Förderung
- Ministry of Trade, Industry and Energy (MOTIE)
- Korea government (MSIT)
- Ministry of Education (MOE, Korea)
Zitieren
APA
Lee, W., Ju, M., Sim, Y., Jung, Y. K., Kim, T. H., & Kim, Y. (2024). Adversarial Attacks on Medical Segmentation Model via Transformation of Feature Statistics. Applied Sciences, 14(6), 2576. https://doi.org/10.3390/app14062576
Vancouver
Lee W, Ju M, Sim Y, Jung YK, Kim TH, Kim Y. Adversarial Attacks on Medical Segmentation Model via Transformation of Feature Statistics. Applied Sciences. 2024;14(6):2576. doi:10.3390/app14062576
BibTeX
@article{Lee2024,
author = {Lee, Woonghee and Ju, Mingeon and Sim, Yura and Jung, Young Kul and Kim, Tae Hyung and Kim, Younghoon},
title = {Adversarial Attacks on Medical Segmentation Model via Transformation of Feature Statistics},
journal = {Applied Sciences},
year = {2024},
volume = {14},
number = {6},
pages = {2576},
doi = {10.3390/app14062576}
}
EUVIMED