SavedAdd your journal

Journals / Erciyes Üniversitesi Fen Bilimleri Enstitüsü Fen Bilimleri Dergisi / 2026 / Cilt: 42 Sayı: 3

Evaluation of Adversarial Robustness in Deep Learning-Based Colorectal Histopathology Classification

Derin Öğrenme Tabanlı Kolorektal Histopatoloji Sınıflandırmasında Saldırılara Karşı Dayanıklılığın Değerlendirilmesi

Tuğba Karataş0009-0001-3396-6157
Mustafa Etcil(ABDULLAH GUL UNIVERSITY)0000-0002-5573-3418
Burak Kolukisa(Kayseri University)0000-0003-0423-4595
Pages
—
Publication Date
2026-10-24
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Both defense strategies generally improved performance under the evaluated adversarial conditions: adversarial training substantially improved performance at ε = 0.01, whereas mixed-ε CAE purification generally showed stronger relative performance at higher perturbation levels.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Background/Aim: Deep learning models can achieve high accuracy in colorectal histopathology classification, but their performance may decrease under adversarial perturbations.
📊
Key Findings & Data:This study aimed to evaluate the adversarial robustness of a 2D-CNN across two colorectal histopathology datasets under FGSM, PGD, and BIM attacks and to compare adversarial training with convolutional autoencoder (CAE)-based input purification as defense strategies.
💡
Conclusion & Impact:Both defense strategies generally improved performance under the evaluated adversarial conditions: adversarial training substantially improved performance at ε = 0.01, whereas mixed-ε CAE purification generally showed stronger relative performance at higher perturbation levels.
🔬Design:Modeling & Simulation
📍Scope:2016 Verileri
⚙️Variables:Kolorektal histopatoloji · Derin öğrenme · Adversarial saldırılar

Abstract

Background/Aim: Deep learning models can achieve high accuracy in colorectal histopathology classification, but their performance may decrease under adversarial perturbations. This study aimed to evaluate the adversarial robustness of a 2D-CNN across two colorectal histopathology datasets under FGSM, PGD, and BIM attacks and to compare adversarial training with convolutional autoencoder (CAE)-based input purification as defense strategies. Methods: Experiments were conducted on the Kather Texture 2016 and CRC-VAL-HE-7K datasets, containing 5,000 and 7,180 image patches from eight and nine colorectal tissue classes, respectively. A 2D-CNN was trained using five-fold stratified cross-validation and evaluated on held-out test sets. White-box FGSM, PGD, and BIM attacks were applied across multiple perturbation levels. Adversarial training was examined using three clean-to-adversarial training compositions. For input purification, shallow, deep, and mixed-ε deep CAE models were evaluated. Performance was assessed using accuracy, macro precision, macro recall, and macro F1-score. Results: The 2D-CNN achieved 87.70% and 89.35% clean test accuracy on Kather Texture 2016 and CRC-VAL-HE-7K, respectively, but performance declined markedly under adversarial attacks, particularly on CRC-VAL-HE-7K. Adversarial training substantially improved robustness at ε = 0.01, although the best training composition varied across datasets and attacks. Mixed-ε deep CAE generally provided the strongest purification performance at higher perturbation levels. Conclusion: Despite its clean-image classification performance, the 2D-CNN remained vulnerable to adversarial perturbations. Both defense strategies generally improved performance under the evaluated adversarial conditions: adversarial training substantially improved performance at ε = 0.01, whereas mixed-ε CAE purification generally showed stronger relative performance at higher perturbation levels.

Özet

Arka Plan/Amaç: Derin öğrenme modelleri kolorektal histopatoloji sınıflandırmasında yüksek doğruluk sağlayabilse de adversarial pertürbasyonlar altında performansları azalabilmektedir. Bu çalışmada, bir 2D-CNN modelinin iki kolorektal histopatoloji veri setinde FGSM, PGD ve BIM saldırıları altındaki adversarial dayanıklılığının değerlendirilmesi ve adversarial eğitim ile evrişimsel otoenkoder (CAE) tabanlı girdi arındırma yöntemlerinin savunma stratejileri olarak karşılaştırılması amaçlanmıştır. Yöntemler: Deneyler, sırasıyla sekiz ve dokuz kolorektal doku sınıfına ait 5.000 ve 7.180 RGB görüntü yaması içeren Kather Texture 2016 ve CRC-VAL-HE-7K veri setleri üzerinde gerçekleştirilmiştir. Bir 2D-CNN modeli beş katlı tabakalı çapraz doğrulama kullanılarak eğitilmiş ve eğitimden ayrı tutulan test setlerinde değerlendirilmiştir. Beyaz kutu FGSM, PGD ve BIM saldırıları birden fazla pertürbasyon düzeyinde uygulanmıştır. Adversarial eğitim, üç farklı temiz-adversarial eğitim verisi bileşimi kullanılarak incelenmiştir. Girdi arındırma için sığ, derin ve karma-ε derin CAE modelleri değerlendirilmiştir. Performans; doğruluk, makro kesinlik, makro duyarlılık ve makro F1-skoru kullanılarak değerlendirilmiştir. Bulgular: 2D-CNN modeli, Kather Texture 2016 ve CRC-VAL-HE-7K veri setlerinde temiz test görüntüleri üzerinde sırasıyla %87,70 ve %89,35 doğruluk elde etmiş; ancak adversarial saldırılar altında, özellikle CRC-VAL-HE-7K veri setinde, performans belirgin şekilde azalmıştır. Adversarial eğitim, ε=0,01 düzeyinde dayanıklılığı önemli ölçüde artırmış olmakla birlikte, en iyi eğitim bileşimi veri setine ve saldırı türüne göre değişiklik göstermiştir. Karma-ε derin CAE modeli ise genel olarak daha yüksek pertürbasyon düzeylerinde en güçlü arındırma performansını sağlamıştır. Sonuç: 2D-CNN, temiz görüntülerdeki sınıflandırma performansına rağmen adversarial pertürbasyonlara karşı savunmasız kalmıştır. Her iki savunma stratejisi de değerlendirilen adversarial koşullar altında genel olarak performansı artırmıştır: adversarial eğitim ε = 0.01 düzeyinde performansı önemli ölçüde iyileştirirken, karma-ε CAE tabanlı girdi arındırma genel olarak daha yüksek pertürbasyon düzeylerinde daha güçlü göreli performans göstermiştir.

🔖

Cite This Article

Karataş, T., Etcil, M., Kolukisa, B. (2026). Evaluation of Adversarial Robustness in Deep Learning-Based Colorectal Histopathology Classification. Erciyes Üniversitesi Fen Bilimleri Enstitüsü Fen Bilimleri Dergisi, Cilt: 42 Sayı: 3. https://doi.org/10.65520/erciyesfen.1997805

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (48)

  1. Hossain, M. S., Karuniawati, H., Jairoun, A. A., Urbi, Z., Ooi, D. J., John, A.,. & Hadi, M. A. (2022). Colorectal cancer: A review of carcinogenesis, global epidemiology, current challenges, risk factors, preventive and treatment strategies. Cancers, 14(7), 1732.
  2. Bray, F., Laversanne, M., Sung, H., Ferlay, J., Siegel, R. L., Soerjomataram, I., et al. (2024). Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer Journal for Clinicians, 74(3), 229–263.
  3. Chhikara, B. S., & Parang, K. (2023). Global cancer statistics 2022: The trends projection analysis. Chemical Biology Letters, 10(1), 451.
  4. Simon, K. (2016). Colorectal cancer development and advances in screening. Clinical Interventions in Aging, 11, 967–976.
  5. Mattiuzzi, C., Sanchis-Gomar, F., & Lippi, G. (2019). Concise update on colorectal cancer epidemiology. Annals of Translational Medicine, 7(21), 609.
  6. Pacal, I., & Attallah, O. (2025). Hybrid deep learning model for automated colorectal cancer detection using local and global feature extraction. Knowledge-Based Systems, 319, 113625.
  7. Pickhardt, P. J., Pooler, B. D., Kim, D. H., Hassan, C., Matkowskyj, K. A., & Halberg, R. B. (2018). The natural history of colorectal polyps: Overview of predictive static and dynamic features. Gastroenterology Clinics of North America, 47(3), 515–536.
  8. Kather, J. N., Krisam, J., Charoentong, P., Luedde, T., Herpel, E., Weis, C. A.,. & Halama, N. (2019). Predicting survival from colorectal cancer histology slides using deep learning: A retrospective multicenter study. PLoS Medicine, 16(1), e1002730.
  9. Kather, J. N., Weis, C. A., Bianconi, F., Melchers, S. M., Schad, L. R., Gaiser, T., et al. (2016). Multi-class texture analysis in colorectal cancer histology. Scientific Reports, 6(1), 27988.
  10. Etcil, M., Dedeturk, B. K., Kolukisa, B., Bakir-Gungor, B., & Gungor, V. C. (2025). Breast cancer detection using a new parallel hybrid logistic regression model trained by particle swarm optimization and clonal selection algorithms. Concurrency and Computation: Practice and Experience, 37(12–14), e70107.
  11. Shafi, A. S. M., Molla, M. I., Jui, J. J., & Rahman, M. M. (2020). Detection of colon cancer based on microarray dataset using machine learning as feature selection and classification techniques. SN Applied Sciences, 2(7), 1243.
  12. Alqudah, A. M., & Alqudah, A. (2022). Improving machine learning recognition of colorectal cancer using 3D GLCM applied to different color spaces. Multimedia Tools and Applications, 81(8), 10839–10860.
  13. Aggarwal, R., Sounderajah, V., Martin, G., Ting, D. S., Karthikesalingam, A., King, D.,. & Darzi, A. (2021). Diagnostic accuracy of deep learning in medical imaging: A systematic review and meta-analysis. npj Digital Medicine, 4(1), 65.
  14. Kolukisa, B., Görmez, Y., & Aydın, Z. (2022). A transfer learning approach for skin cancer subtype detection. In The International Conference on Artificial Intelligence and Applied Mathematics in Engineering (pp. 337–347). Springer International Publishing.
  15. Kolukisa, B. (2026). Pathogenicity assessment of breast cancer-associated BRCA1/2 variants using an explainable machine learning approach. Journal of the Faculty of Engineering and Architecture of Gazi University, 41(2), 1117–1128.
  16. Acs, B., Rantalainen, M., & Hartman, J. (2020). Artificial intelligence as the next step towards precision pathology. Journal of Internal Medicine, 288(1), 62–81.
  17. Chlorogiannis, D. D., Verras, G. I., Tzelepi, V., Chlorogiannis, A., Apostolos, A., Kotis, K.,. & Mulita, F. (2023). Tissue classification and diagnosis of colorectal cancer histopathology images using deep learning algorithms: Is the time ripe for clinical practice implementation? Gastroenterology Review/Przegląd Gastroenterologiczny, 18(4), 353–367.
  18. Kumar, A., Vishwakarma, A., & Bajaj, V. (2023). CRCCN-Net: Automated framework for classification of colorectal tissue using histopathological images. Biomedical Signal Processing and Control, 79, 104172.
  19. Albashish, D. (2022). Ensemble of adapted convolutional neural network methods for classifying colon histopathological images. PeerJ Computer Science, 8, e1031.
  20. Dabass, M., Vashisth, S., & Vig, R. (2022). A convolutional neural network with multi-level convolutional and attention learning for classification of cancer grades and tissue structures in colon histopathological images. Computers in Biology and Medicine, 147, 105680.
  21. Wang, X., Yang, S., Zhang, J., Wang, M., Zhang, J., Yang, W.,. & Han, X. (2022). Transformer-based unsupervised contrastive learning for histopathological image classification. Medical image analysis, 81, 102559.
  22. Chen, R. J., Ding, T., Lu, M. Y., Williamson, D. F., Jaume, G., Song, A. H.,. & Mahmood, F. (2024). Towards a general-purpose foundation model for computational pathology. Nature medicine, 30(3), 850-862.
  23. Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I. J., & Fergus, R. (2014). Intriguing properties of neural networks. In International Conference on Learning Representations.
  24. Carlini, N., & Wagner, D. (2017). Adversarial examples are not easily detected: Bypassing ten detection methods. In Proceedings of the 10th ACM Workshop on Artificial Intelligence and Security (pp. 3–14).
  25. Goodfellow, I. J., Shlens, J., & Szegedy, C. (2014). Explaining and harnessing adversarial examples. arXiv Preprint, arXiv:1412.6572.
  26. Croce, F., & Hein, M. (2020, November). Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks. In International conference on machine learning (pp. 2206-2216). PMLR.
  27. Finlayson, S. G., Bowers, J. D., Ito, J., Zittrain, J. L., Beam, A. L., & Kohane, I. S. (2019). Adversarial attacks on medical machine learning. Science, 363(6433), 1287–1289.
  28. Ma, X., Niu, Y., Gu, L., Wang, Y., Zhao, Y., Bailey, J., & Lu, F. (2021). Understanding adversarial attacks on deep learning-based medical image analysis systems. Pattern Recognition, 110, 107332.
  29. Paschali, M., Conjeti, S., Navarro, F., & Navab, N. (2018). Generalizability versus robustness: Investigating medical imaging networks using adversarial examples. In International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 493–501). Springer International Publishing.
  30. Dong, J., Chen, J., Xie, X., Lai, J., & Chen, H. (2024). Survey on adversarial attack and defense for medical image analysis: Methods and challenges. ACM Computing Surveys, 57(3), 1–38.
  31. Ghaffari Laleh, N., Truhn, D., Veldhuizen, G. P., Han, T., van Treeck, M., Buelow, R. D.,. & Kather, J. N. (2022). Adversarial attacks and adversarial robustness in computational pathology. Nature Communications, 13(1), 5711.
  32. Thota, P., Veerla, J. P., Guttikonda, P. S., Nasr, M. S., Nilizadeh, S., & Luber, J. M. (2024). Demonstration of an adversarial attack against a multimodal vision-language model for pathology imaging. In 2024 IEEE International Symposium on Biomedical Imaging (pp. 1–5). IEEE.
  33. Mehrtens, H. A., Kurz, A., Bucher, T. C., & Brinker, T. J. (2023). Benchmarking common uncertainty estimation methods with histopathological images under domain shift and label noise. Medical image analysis, 89, 102914.
  34. Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2017). Towards deep learning models resistant to adversarial attacks. arXiv Preprint, arXiv:1706.06083.
  35. Suciati, N., & Yuniarti, A. (2024). Adversarial Training For Robust Defense In Cnn Models For Lung And Colon Histopathological Images. IPTEK The Journal for Technology and Science, 35(2), 135-143.
  36. Meng, D., & Chen, H. (2017). MagNet: A two-pronged defense against adversarial examples. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security (pp. 135–147).
  37. Shim, J., & Jo, K. (2025). Siamese denoising autoencoders for enhancing adversarial robustness in medical image analysis. IEEE Access, 13, 86333-86343.
  38. LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278–2324.
  39. Musthafa, M. M., TR, M., V, V. K., & Guluwadi, S. (2024). Enhanced skin cancer diagnosis using optimized CNN architecture and checkpoints for automated dermatological lesion classification. BMC Medical Imaging, 24(1), 201.
  40. Manzoor, A., Fargetta, G., Ortis, A., & Battiato, S. (2026). ShieldNet: A novel adversarially resilient convolutional neural network for robust image classification. Applied Sciences, 16(3), 1254.
  41. Kurakin, A., Goodfellow, I., & Bengio, S. (2016). Adversarial machine learning at scale. arXiv preprint arXiv:1611.01236.
  42. Hossain, M. S., Karuniawati, H., Jairoun, A. A., Urbi, Z., Ooi, D. J., John, A.,. & Hadi, M. A. (2022). Colorectal cancer: A review of carcinogenesis, global epidemiology, current challenges, risk factors, preventive and treatment strategies. Cancers, 14(7), 1732.
  43. Kather, J. N., Krisam, J., Charoentong, P., Luedde, T., Herpel, E., Weis, C. A.,. & Halama, N. (2019). Predicting survival from colorectal cancer histology slides using deep learning: A retrospective multicenter study. PLoS Medicine, 16(1), e1002730.
  44. Aggarwal, R., Sounderajah, V., Martin, G., Ting, D. S., Karthikesalingam, A., King, D.,. & Darzi, A. (2021). Diagnostic accuracy of deep learning in medical imaging: A systematic review and meta-analysis. npj Digital Medicine, 4(1), 65.
  45. Chlorogiannis, D. D., Verras, G. I., Tzelepi, V., Chlorogiannis, A., Apostolos, A., Kotis, K.,. & Mulita, F. (2023). Tissue classification and diagnosis of colorectal cancer histopathology images using deep learning algorithms: Is the time ripe for clinical practice implementation? Gastroenterology Review/Przegląd Gastroenterologiczny, 18(4), 353–367.
  46. Wang, X., Yang, S., Zhang, J., Wang, M., Zhang, J., Yang, W.,. & Han, X. (2022). Transformer-based unsupervised contrastive learning for histopathological image classification. Medical image analysis, 81, 102559.
  47. Chen, R. J., Ding, T., Lu, M. Y., Williamson, D. F., Jaume, G., Song, A. H.,. & Mahmood, F. (2024). Towards a general-purpose foundation model for computational pathology. Nature medicine, 30(3), 850-862.
  48. Ghaffari Laleh, N., Truhn, D., Veldhuizen, G. P., Han, T., van Treeck, M., Buelow, R. D.,. & Kather, J. N. (2022). Adversarial attacks and adversarial robustness in computational pathology. Nature Communications, 13(1), 5711.