Exploring Fractal Recurrence Representations with CNNs for Oral Epithelial Dysplasia Classification

• Vitória F. C. Silva UFU
• Gustavo C. Miranda UFU
• Daniel B. Gonçalves UFU
• Domingos L. L. de Oliveira UFU / IFSP
• Guilherme F. Roberto UNESP
• Leandro A. Neves UNESP
• Adriano B. Silva UFU
• Marcelo Z. do Nascimento UFU

DOI: https://doi.org/10.5753/sbcas.2026.20798

Abstract

Early diagnosis of oral epithelial dysplasia is crucial, yet manual histopathological evaluation is influenced by subjective aspects and observer variability. Computer-aided diagnosis (CAD) systems have emerged as a strategy to assist in this analysis. However, traditional neural network approaches often entail high computational costs and fail to explore heterogeneous ensemble strategies to improve generalization. This paper presents a comparative performance analysis of lightweight convolutional neural network (CNN) architectures (MobileNetV2 and EfficientNet-B0) for classifying oral dysplasia using the OralEpitheliumDB dataset. The methodology contrasts the use of original histological images against Recurrence Plot (RP) representations generated from multiscale fractal descriptors. Experimental results demonstrated that the RP transformation yielded substantial improvements across all evaluated metrics. Notably, the heterogeneous ensemble combining MobileNet and EfficientNet, both trained on the RP representation, achieved 100% accuracy, sensitivity, and specificity. In conclusion, recurrence analysis reveals latent texture patterns not captured in the original spatial domain, enabling the implementation of robust and computationally efficient diagnostic screening systems.

References

Adel, D., Mounir, J., El-Shafey, M., Eldin, Y. A., El Masry, N., AbdelRaouf, A., and Abd Elhamid, I. S. (2018). Oral epithelial dysplasia computer aided diagnostic approach. In 2018 13th International Conference on Computer Engineering and Systems (ICCES), pages 313–318. IEEE.

Alajaji, S. A., Khoury, Z. H., Jessri, M., Sciubba, J. J., and Sultan, A. S. (2024). An update on the use of artificial intelligence in digital pathology for oral epithelial dysplasia research. Head and neck pathology, 18(1):38.

Cheung, V. K., Hulme, K., Schifter, M., Palme, C., Low, T.-H. H., Clark, J., and Gupta, R. (2022). Oral epithelial dysplasia: a review of diagnostic criteria for anatomic pathologists. Advances in Anatomic Pathology, 29(4):227–240.

Dabeer, S., Khan, M. M., and Islam, S. (2019). Cancer diagnosis in histopathological image: Cnn based approach. Informatics in Medicine Unlocked, 16:100231.

Deif, M. A., Attar, H., Amer, A., Elhaty, I. A., Khosravi, M. R., Solyman, A. A., et al. (2022). Diagnosis of oral squamous cell carcinoma using deep neural networks and binary particle swarm optimization on histopathological images: an aiomt approach. Computational Intelligence and Neuroscience, 2022.

Eckmann, J.-P., Kamphorst, S. O., Ruelle, D., et al. (1995). Recurrence plots of dynamical systems. World Scientific Series on Nonlinear Science Series A, 16:441–446.

Longo, L. H. d. C., Roberto, G. F., Tosta, T. A., de Faria, P. R., Loyola, A. M., Cardoso, S. V., Silva, A. B., do Nascimento, M. Z., and Neves, L. A. (2023). Classification of multiple h&e images via an ensemble computational scheme. Entropy, 26(1):34.

Müller, S. (2018). Oral epithelial dysplasia, atypical verrucous lesions and oral potentially malignant disorders: focus on histopathology. Oral surgery, oral medicine, oral pathology and oral radiology, 125(6):591–602.

Pereira, D. C., Silva, A. B., Longo, L. H., Loyola, A. M., Cardoso, S. V., de Faria, P. R., Tosta, T. A., Neves, L. A., Martins, A. S., and Nascimento, M. Z. (2026). Evaluation of fractal descriptors, deep features and xai representations with lasso-regularized hermite polynomial classifier for h&e histological image classification. Biomedical Signal Processing and Control, 112:108393.

Pham, T. D. (2021). From raw pixels to recurrence image for deep learning of benign and malignant mediastinal lymph nodes on computed tomography. IEEE Access, 9:96267–96278.

Pimenta-Barros, L. A., Ramos-García, P., González-Moles, M. Á., Aguirre-Urizar, J. M., and Warnakulasuriya, S. (2025). Malignant transformation of oral leukoplakia: Systematic review and comprehensive meta-analysis. Oral diseases, 31(1):69–80.

Roberto, G. F. et al. (2021). Associação entre atributos manuais e aprendizado profundo baseada em geometria fractal para classificação de imagens histológicas.

Roberto, G. F., Neves, L. A., Lumini, A., Martins, A. S., and Nascimento, M. Z. d. (2024). An ensemble of learned features and reshaping of fractal geometry-based descriptors for classification of histological images. Pattern Analysis and Applications, 27(1):8.

Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., and Chen, L.-C. (2018). Mobilenetv2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

Santos, M. d. O. (2023). Estimativa de incidência de câncer no brasil, 2023-2025. Revista Brasileira de Cancerologia, 69(1).

Silva, A. B., Martins, A. S., Tosta, T. A. A., Loyola, A. M., Cardoso, S. V., Neves, L. A., de Faria, P. R., and do Nascimento, M. Z. (2024). Oralepitheliumdb: A dataset for oral epithelial dysplasia image segmentation and classification. Journal of Imaging Informatics in Medicine, 37(4):1691–1710.

Silva, A. B., Martins, A. S., Tosta, T. A. A., Neves, L. A., Servato, J. P. S., de Araújo, M. S., de Faria, P. R., and do Nascimento, M. Z. (2022). Computational analysis of histological images from hematoxylin and eosin-stained oral epithelial dysplasia tissue sections. Expert Systems with Applications, 193:116456.

Tan, M. and Le, Q. V. (2019). Efficientnet: Rethinking model scaling for convolutional neural networks. In Proceedings of the International Conference on Machine Learning (ICML).

van Zelst, J. C., Tan, T., Mann, R. M., and Karssemeijer, N. (2020). Validation of radiologists’ findings by computer-aided detection (cad) software in breast cancer detection with automated 3d breast ultrasound: a concept study in implementation of artificial intelligence software. Acta Radiologica, 61(3):312–320.

Warnakulasuriya, S., Reibel, J., Bouquot, J., and Dabelsteen, E. (2008). Oral epithelial dysplasia classification systems: predictive value, utility, weaknesses and scope for improvement. Journal of oral pathology & medicine, 37(3):127–133.