EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Analyzing CNN-Based Glaucoma Decision Criteria Using Adversarial Examples

Shinichiro Ishikawa, Hiyori Sakemi, Koki Hirose, Tahsina Nabiha Khan, Kenshin Mizoe, Ikki Osaka, Osamu Fukuda, Nobuhiko Yamaguchi, Masateru Kawakubo, Hiroshi Okumura

Technologies · 16.07.2026 · 14 (7) · 435

DOI: 10.3390/technologies14070435 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Technologies, online erschienen am 16.07.2026. Verfasst von Shinichiro Ishikawa und 9 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Japan Society for the Promotion of Science.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Glaucoma is a leading cause of blindness, and early detection is critical. Convolutional neural networks (CNNs) have shown impressive performance in glaucoma diagnosis, but their black-box nature remains a barrier to clinical use. Existing explainable AI (XAI) methods such as Grad-CAM have limitations in identifying and quantifying subtle regional features. In this study, we propose a method to clarify what CNNs focus on by analyzing how model performance changes under localized adversarial noise. Using VGG16 for glaucoma classification, we applied noise generated by the Fast Gradient Sign Method (FGSM) to the whole fundus image and to specific subregions, then compared the impact on classification performance. Results showed that perturbations to the optic disc, especially its outer margin, had the greatest effect on model performance. This suggests that the CNN captures fine anatomical features such as optic disc cupping and neuroretinal rim thinning, which aligns with what ophthalmologists typically look for. At the same time, perturbations in the macula and perivascular regions also affected performance, indicating gaps between current clinical diagnostic criteria and the CNN’s decision-making process. This approach can help establish the clinical reliability of CNNs and may also reveal features that have not been recognized in conventional clinical practice.

Förderung

  • Japan Society for the Promotion of Science

Zitieren

APA

Ishikawa, S., Sakemi, H., Hirose, K., Khan, T. N., Mizoe, K., Osaka, I., Fukuda, O., Yamaguchi, N., Kawakubo, M., & Okumura, H. (2026). Analyzing CNN-Based Glaucoma Decision Criteria Using Adversarial Examples. Technologies, 14(7), 435. https://doi.org/10.3390/technologies14070435

Vancouver

Ishikawa S, Sakemi H, Hirose K, Khan TN, Mizoe K, Osaka I, et al. Analyzing CNN-Based Glaucoma Decision Criteria Using Adversarial Examples. Technologies. 2026;14(7):435. doi:10.3390/technologies14070435

BibTeX
@article{Ishikawa2026,
  author = {Ishikawa, Shinichiro and Sakemi, Hiyori and Hirose, Koki and Khan, Tahsina Nabiha and Mizoe, Kenshin and Osaka, Ikki and Fukuda, Osamu and Yamaguchi, Nobuhiko and Kawakubo, Masateru and Okumura, Hiroshi},
  title = {Analyzing CNN-Based Glaucoma Decision Criteria Using Adversarial Examples},
  journal = {Technologies},
  year = {2026},
  volume = {14},
  number = {7},
  pages = {435},
  doi = {10.3390/technologies14070435}
}
Kostenlos starten →