EUVIMEDEUVIMED entdecken
Crossref

Medical VLP Model Is Vulnerable: Toward Multimodal Adversarial Attack on Large Medical Vision-Language Models

Zimu Lu, Ning Xu, Hongshuo Tian, Lanjun Wang, An-An Liu

IEEE Transactions on Circuits and Systems for Video Technology · 2026-02 · 36 (2) · 2478-2491

DOI: 10.1109/tcsvt.2025.3602970 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Circuits and Systems for Video Technology, 2026-02. Verfasst von Zimu Lu und 4 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 86 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Medical Visual Question Answering (Medical VQA) is an essential task that facilitates the automated interpretation of complex clinical imagery with corresponding textual questions, thereby supporting both clinicians and patients in making informed medical decisions. With the rapid progress of Vision-Language Pretraining (VLP) in general domains, the development of medical VLP models has emerged as a rapidly growing interdisciplinary area at the intersection of artificial intelligence (AI) and healthcare. However, few works have been proposed to evaluate the adversarial robustness of medical VLP models, which faces two primary challenges: 1) the complexity of medical texts, stemming from the presence of terminologies, poses significant challenges for models in comprehending the text for adversarial attack; 2) the diversity of medical images arises from the variety of anatomical regions depicted, which requires models to determine critical anatomical regions for attack. In this paper, we propose a novel multimodal adversarial attack generator for evaluating the robustness of medical VLP models. Specifically, for the complexity of medical texts, we integrate medical knowledge when crafting text adversarial samples, which can facilitate the terminologies understanding and adversarial strength; for the diversity of medical images, we divide the anatomical regions into either global or local regions in medical images, which are determined by learned balance weights for perturbations. Our experimental study not only provides a quantitative understanding in medical VLP models, but also underscores the critical need for thorough safety evaluations before implementing them in real-world medical applications.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Lu, Z., Xu, N., Tian, H., Wang, L., & Liu, A. (2026). Medical VLP Model Is Vulnerable: Toward Multimodal Adversarial Attack on Large Medical Vision-Language Models. IEEE Transactions on Circuits and Systems for Video Technology, 36(2), 2478-2491. https://doi.org/10.1109/tcsvt.2025.3602970

Vancouver

Lu Z, Xu N, Tian H, Wang L, Liu A. Medical VLP Model Is Vulnerable: Toward Multimodal Adversarial Attack on Large Medical Vision-Language Models. IEEE Transactions on Circuits and Systems for Video Technology. 2026;36(2):2478-2491. doi:10.1109/tcsvt.2025.3602970

BibTeX
@article{Lu2026,
  author = {Lu, Zimu and Xu, Ning and Tian, Hongshuo and Wang, Lanjun and Liu, An-An},
  title = {Medical VLP Model Is Vulnerable: Toward Multimodal Adversarial Attack on Large Medical Vision-Language Models},
  journal = {IEEE Transactions on Circuits and Systems for Video Technology},
  year = {2026},
  volume = {36},
  number = {2},
  pages = {2478--2491},
  doi = {10.1109/tcsvt.2025.3602970}
}
Kostenlos starten →