Evaluation of Machine Learning Methods for Oral Cavity Histopathological Cancer Classification in a Brazilian Cohort

• Matheus de Freitas Oliveira Baffa USP
• Luciano Bachmann USP
• Denise Maria Zezell IPEN
• Leandro Luongo Matos USP
• Joaquim Cezar Felipe USP

DOI: https://doi.org/10.5753/sbcas.2026.20718

Resumo

Histopathological evaluation is the gold standard for oral cavity cancer diagnosis, but it is time-consuming and subject to inter-observer variability. In this study, we compare radiomics, convolutional neural networks, and a DINOv2-based transformer model for histopathological image classification using a Brazilian cohort. Radiomic features combined with a fully-connected neural network achieved the best overall performance, reaching 93.18% accuracy, 95.56% sensitivity, and 92.50% specificity, while requiring lower computational cost than end-to-end deep learning models. These findings suggest that radiomics provides an efficient alternative for oral cavity cancer classification, particularly in scenarios with limited data and computational resources.

Referências

Asif, S., Wang, V. Y., and Xu, D. (2025). Oraltransnet: A novel hybrid model integrating transformer attention and cnn features for accurate diagnosis of mouth and oral diseases. Engineering Applications of Artificial Intelligence, 159:111609.

Asnake, N. W., Ayalew, A. M., and Engda, A. A. (2025). Detection of oral squamous cell carcinoma cancer using alexnet on histopathological images. Discover Applied Sciences, 7:155.

Das, N., Hussain, E., and Mahanta, L. B. (2020). Automated classification of cells into multiple classes in epithelial tissue of oral squamous cell carcinoma using transfer learning and convolutional neural network. Neural Networks, 128:47–60.

Deo, B. S. et al. (2024). Supremacy of attention-based transformer in oral cancer classification using histopathology images. International Journal of Data Science and Analytics, pages 1–19.

Folmsbee, J., Liu, X., Brandwein-Weber, M., and Doyle, S. (2018). Active deep learning: Improved training efficiency of convolutional neural networks for tissue classification in oral cavity cancer. In 2018 IEEE 15th international symposium on biomedical imaging (ISBI 2018), pages 770–773. IEEE.

Komura, D., Ochi, M., and Ishikawa, S. (2024). Machine learning methods for histopathological image analysis: Updates in 2024. Computational and Structural Biotechnology Journal, 27:383–400.

Lu, M. Y. et al. (2021). Data-efficient and weakly supervised computational pathology on whole-slide images. Nature biomedical engineering, 5(6):555–570.

Maran, B., de Brot, L., Corassa, M., and de Paula, R. (2022). Disagreement in anatomopathological review reports in gynaecological pathology and its impact on treatment: the importance of the subspecialist pathologist in a cancer centre. In VIRCHOWS ARCHIV, volume 481, pages S115–S115. SPRINGER ONE.

Matos, L. L. et al. (2020). Cancer-associated fibroblast regulation by micrornas promotes invasion of oral squamous cell carcinoma. Oral Oncology, 110:104909.

Mayo Clinic (2023). Biopsy: Types of biopsy procedures used to diagnose cancer. Available at [link]. Accessed:on 25 Feb. 2025.

Menderico Junior, G. M. et al. (2021). Microrna-mediated extracellular matrix remodeling in squamous cell carcinoma of the oral cavity. Head & Neck, 43(8):2364–2376.

National Cancer Institute (2021). Head and neck cancers: Fact sheet. Available at [link]. Accessed on 27 Feb. 2026.

National Health Service (2021). Head and neck cancer. Available at [link]. Accessed on 27 Feb. 2026.

Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al. (2023). Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193.

Pham, T. D. (2025). Integrating support vector machines and deep learning features for oral cancer histopathology analysis. Biology Methods and Protocols, 10(1):bpaf034.

Ramani, R. S., Tan, I., Bussau, L., et al. (2025). Convolutional neural networks for accurate real-time diagnosis of oral epithelial dysplasia and oral squamous cell carcinoma using high-resolution in vivo confocal microscopy. Scientific Reports, 15:2555.

Tan, Y. et al. (2023). Oral squamous cell carcinomas: state of the field and emerging directions. International journal of oral science, 15(1):44.

Tomaszewski, M. R. and Gillies, R. J. (2021). The biological meaning of radiomic features. Radiology, 298(3):505–516.

Wang, X. et al. (2021). Transpath: Transformer-based self-supervised learning for histopathological image classification. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 186–195. Springer.

Warin, K., Limprasert, W., Suebnukarn, S., Jinaporntham, S., and Jantana, P. (2021). Automatic classification and detection of oral cancer in photographic images using deep learning algorithms. Journal of Oral Pathology & Medicine, 50(9):911–918.