Grad-CAM Sensitivity to Orientation Transformations in Breast Cancer Histopathological Images

Felipe Moura da Silva; Roberta Barbosa Oliveira

doi:10.22456/2175-2745.150979

Authors

Felipe Moura da Silva Universidade de Brasília (UnB) https://orcid.org/0009-0007-4627-118X
Roberta Barbosa Oliveira Universidade de Brasília (UnB)

DOI:

https://doi.org/10.22456/2175-2745.150979

Keywords:

artificial intelligence, computer vision, explainable artificial intelligence, saliency maps

Abstract

Breast cancer is the second most prevalent among women, and early diagnosis is essential to reduce mortality. Machine learning models are being used to improve diagnostic accuracy, especially through biopsies. However, the growing complexity of neural networks has raised concerns about result interpretability, driving the adoption of explainability methods such as the Grad-CAM, which generates saliency maps in CNN models. This paper investigates the sensitivity of Grad-CAM to changes in orientation and position in breast histopathological images. CNN's ResNet50, VGG16, and Xception were evaluated, trained with the BreakHis dataset, which comprises images at different magnifications. Test samples were modified by rotations and flips to assess the robustness of both the models and the explanations. Xception achieved the best results in terms of accuracy (94.47%) and F1-score (95.85%) with 100X images, although the explanatory maps varied according to the applied transformation.

Downloads

Download data is not yet available.

References

[1] MARMITT, N. et al. Cenário das publicações sobre perfil epidemiológico e clínico do câncer de mama: Revisão de escopo. LUMEN ET VIRTUS, v. 16, n. 48, p. 6027–6048, maio 2025. ISSN 2177-2789. Disponível em: ⟨https://periodicos.newsciencepubl.com/LEV/article/view/5511⟩.

[2] RAMALHO, R. B. et al. Diagnóstico e tratamento do câncer de mama: uma revisão de literatura. Brazilian Journal of Implantology and Health Sciences, v. 6, n. 8, p. 1040–1050, ago. 2024. Disponível em: ⟨https://bjihs.emnuvens.com.br/bjihs/article/view/2866⟩.

[3] HOUSSEIN, E. H. et al. Deep and machine learning techniques for medical imaging-based breast cancer: A comprehensive review. Expert Systems with Applications, v. 167, p. 114161, abr. 2021. ISSN 0957-4174. Disponível em: ⟨https://linkinghub.elsevier.com/retrieve/pii/S0957417420309015⟩.

[4] YAN, R. et al. Breast cancer histopathological image classification using a hybrid deep neural network. Methods, v. 173, p. 52–60, fev. 2020. ISSN 1046-2023. Disponível em: ⟨https://linkinghub.elsevier.com/retrieve/pii/S1046202319300349⟩.

[5] ČERNEVIČIENĖ, J.; KABAŠINSKAS, A. Explainable artificial intelligence (XAI) in finance: a systematic literature review. Artificial Intelligence Review, v. 57, n. 8, p. 216, jul. 2024. ISSN 1573-7462. Disponível em: ⟨https://doi.org/10.1007/s10462-024-10854-8⟩.

[6] YANG, G.; YE, Q.; XIA, J. Unbox the black-box for the medical explainable ai via multi-modal and multi-centre data fusion: A mini-review, two showcases and beyond. Information Fusion, v. 77, p. 29–52, 2022. ISSN 1566-2535. Disponível em: ⟨https://www.sciencedirect.com/science/article/pii/S1566253521001597⟩.

[7] ARRIETA, A. B. et al. Explainable artificial intelligence (xai): Concepts, taxonomies, opportunities and challenges toward responsible ai. arXiv preprint arXiv:1910.10045, 2019. Disponível em: ⟨https://doi.org/10.48550/arXiv.1910.10045⟩.

[8] ZHAO, X. et al. A review of convolutional neural networks in computer vision. Artificial Intelligence Review, v. 57, n. 4, p. 99, mar. 2024. ISSN 1573-7462. Disponível em: ⟨https://doi.org/10.1007/s10462-024-10721-6⟩.

[9] ESTEVA, A. et al. Deep learning-enabled medical computer vision. npj Digital Medicine, v. 4, n. 1, p. 1–9, jan. 2021. ISSN 2398-6352. Disponível em: ⟨https://www.nature.com/articles/s41746-020-00376-2⟩.

[10] PATRÍCIO, C.; NEVES, J. C.; TEIXEIRA, L. F. Explainable Deep Learning Methods in Medical Image Classification: A Survey. ACM Computing Surveys, v. 56, n. 4, p. 1–41, abr. 2024. ISSN 0360-0300, 1557-7341. Disponível em: ⟨https://dl.acm.org/doi/10.1145/3625287⟩.

[11] AMORIM, J. P. et al. Interpreting Deep Machine Learning Models: An Easy Guide for Oncologists. IEEE Reviews in Biomedical Engineering, v. 16, p. 192–207, 2023. ISSN 1941-1189. Disponível em: ⟨https://ieeexplore.ieee.org/document/9629296⟩.

[12] LIU, Y. et al. Optimized Dropkey-Based Grad-CAM: Toward Accurate Image Feature Localization. Sensors, v. 23, n. 20, p. 8351, jan. 2023. ISSN 1424-8220. Disponível em: ⟨https://www.mdpi.com/1424-8220/23/20/8351⟩.

[13] SELVARAJU, R. R. et al. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. In: 2017 IEEE International Conference on Computer Vision (ICCV). [s.n.], 2017. p. 618–626. Disponível em: ⟨https://ieeexplore.ieee.org/document/8237336⟩.

[14] BAI, S. et al. Breast Cancer Diagnosis: A Comprehensive Exploration of Explainable Artificial Intelligence (XAI) Techniques. arXiv, 2024. Disponível em: ⟨http://arxiv.org/abs/2406.00532⟩.

[15] MALEKI, A.; RAAHEMI, M.; NASIRI, H. Breast cancer diagnosis from histopathology images using deep neural network and XGBoost. Biomedical Signal Processing and Control, v. 86, p. 105152, set. 2023. ISSN 1746-8094. Disponível em: ⟨https://linkinghub.elsevier.com/retrieve/pii/S1746809423005852⟩.

[16] KOSHY, S. S.; ANBARASI, L. J. LMHistNet: Levenberg–Marquardt Based Deep Neural Network for Classification of Breast Cancer Histopathological Images. IEEE Access, v. 12, p. 52051–52066, 2024. ISSN 2169-3536. Disponível em: ⟨https://ieeexplore.ieee.org/document/10491266/⟩.

[17] LIU, S.; HIMEL, G. M. S.; WANG, J. Breast Cancer Classification With Enhanced Interpretability: DALAResNet50 and DT Grad-CAM. IEEE Access, v. 12, p. 196647–196659, 2024. ISSN 2169-3536. Disponível em: ⟨https://ieeexplore.ieee.org/document/10810414/?arnumber=10810414⟩.

[18] AMORIM, J. P. et al. Evaluating the faithfulness of saliency maps in explaining deep learning models using realistic perturbations. Information Processing & Management, v. 60, n. 2, p. 103225, mar. 2023. ISSN 0306-4573. Disponível em: ⟨https://linkinghub.elsevier.com/retrieve/pii/S0306457322003260⟩.

[19] SPANHOL, F. A. et al. A Dataset for Breast Cancer Histopathological Image Classification. IEEE Transactions on Biomedical Engineering, v. 63, n. 7, p. 1455–1462, jul. 2016. ISSN 0018-9294, 1558-2531. Disponível em: ⟨http://ieeexplore.ieee.org/document/7312934/⟩.

[20] HE, K. et al. Deep Residual Learning for Image Recognition. arXiv, 2015. Disponível em: ⟨http://arxiv.org/abs/1512.03385⟩.

[21] SIMONYAN, K.; ZISSERMAN, A. Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv, 2015. Disponível em: ⟨http://arxiv.org/abs/1409.1556⟩.

[22] CHOLLET, F. Xception: Deep Learning with Depthwise Separable Convolutions. arXiv, 2017. Disponível em: ⟨http://arxiv.org/abs/1610.02357⟩.

[23] AL., A. et. Breast cancer detection and classification using deep learning xception algorithm. International Journal of Advanced Computer Science and Applications, The Science and Information Organization, v. 13, n. 7, 2022. Disponível em: ⟨https://dx.doi.org/10.14569/IJACSA.2022.0130729⟩.

[24] BREAST Cancer Histopathological Database (BreakHis) – Laboratório Visão Robótica e Imagem. Disponível em: ⟨https://web.inf.ufpr.br/vri/databases/breast-cancer-histopathological-database-breakhis/⟩.

[25] Cambridge in Colour. Digital Image Interpolation. 2020. Disponível em: ⟨https://www.cambridgeincolour.com/tutorials/image-interpolation.htm⟩.

[26] DENG, J. et al. Imagenet: A large-scale hierarchical image database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition. [S.l.: s.n.], 2009. p. 248–255.

[27] FRAZIER, P. I. A Tutorial on Bayesian Optimization. 2018. Disponível em: ⟨https://arxiv.org/abs/1807.02811⟩.

[28] PRECHELT, L. Early stopping — but when? In: ORR, G. B.; MÜLLER, K.-R. (Ed.). Neural Networks: Tricks of the Trade. Berlin, Heidelberg: Springer, 1998, (Lecture Notes in Computer Science, v. 1524). p. 55–69. Disponível em: ⟨https://doi.org/10.1007/3-540-49430-8_3⟩.

[29] BYLINSKII, Z. et al. What do different evaluation metrics tell us about saliency models? arXiv, 2017. Disponível em: ⟨http://arxiv.org/abs/1604.03605⟩.

[30] LOPATECKI, J. KL Divergence: When To Use Kullback-Leibler divergence. Disponível em: ⟨https://arize.com/blog-course/kl-divergence/⟩.

[31] CHATTOPADHYAY, A. et al. Grad-cam++: Generalized gradient-based visual explanations for deep convolutional networks. In: Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV). [s.n.], 2018. p. 839–847. Disponível em: ⟨https://arxiv.org/abs/1710.11063⟩.

[32] ZEILER, M. D.; FERGUS, R. Visualizing and understanding convolutional networks. In: European Conference on Computer Vision (ECCV). Springer, 2014. p. 818–833. Disponível em: ⟨https://arxiv.org/abs/1311.2901⟩.

[33] SUNDARARAJAN, M.; TALY, A.; YAN, Q. Axiomatic attribution for deep networks. In: Proceedings of the 34th International Conference on Machine Learning (ICML). [s.n.], 2017. p. 3319–3328. Disponível em: ⟨https://arxiv.org/abs/1703.01365⟩.

[34] SPRINGENBERG, J. T. et al. Striving for Simplicity: The All Convolutional Net. 2015. Disponível em: ⟨https://arxiv.org/abs/1412.6806⟩.

[35] POLI, J.-P.; OUERDANE, W.; PIERRARD, R. Generation of textual explanations in xai: the case of semantic annotation. In: 2021 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE). [S.l.: s.n.], 2021. p. 1–6.

[36] SILVA, F. M. Explicabilidade Visual de Redes Convolucionais em Diagnóstico Histopatológico de Câncer de Mama: Sensibilidade do Grad-CAM a Transformações na Orientação das Imagens - Trabalho de Conclusão de Curso - Computação (licenciatura) - Departamento de Ciência da Computação - Universidade de Brasília. 2025.