DenseNet Networks with Multiple Attention Mechanisms Applied to Automatic Penile Cancer Classification in Histopathological Images

• João Guilherme Araujo do Vale UFMA
• Italo Francyles Santos da Silva UFMA
• Caio Eduardo Falcão Matos UFMA
• Geraldo Braz Júnior UFMA
• Marcos Gabriel Mendes Lauande UFMA

DOI: https://doi.org/10.5753/sbcas.2024.2755

Abstract

Penile carcinoma, also known as penile cancer, is a malignant neoplasm that predominantly affects men over 50 years old. Late diagnosis leads to significant complications that deeply impact patients’ lives in psychological, emotional, and social terms. Exams such as histopathological analysis are essential for early diagnosis but require time and highly qualified professionals. The use of deep machine learning, through convolutional neural networks, emerges as a highly applicable approach in this context.In this work, two methods were proposed for the automatic classification of penile cancer, using DenseNet neural networks with attention mechanisms in their architecture. The dataset was provided by the Penile Cancer Project of the Legal Amazon, containing 194 images at 40x and 100x magnifications. As a result, the best model achievedan F1-Score of 93.1% at the 100x magnification.

References

Belfort, F., Silva, I., Silva, A., and Paiva, A. (2023). Detecção de câncer peniano em imagens histopatológicas usando redes neurais convolucionais em cascata. In Anais do XXIII Simpósio Brasileiro de Computação Aplicada à Saúde, pages 328–339, Porto Alegre, RS, Brasil. SBC.

Erickson, B. J., Korfiatis, P., Akkus, Z., and Kline, T. L. (2017). Machine learning for medical imaging. Radiographics, 37(2):505–515.

Huang, G., Liu, Z., Van Der Maaten, L., and Weinberger, K. Q. (2017). Densely connected convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4700–4708.

INCA (2022). Câncer de pênis: Instituto nacional do câncer inca. Acesso em 13/05/2024.

Joseph, F. J. J., Nonsiri, S., and Monsakul, A. (2021). Keras and tensorflow: A hands-on experience. Advanced deep learning for engineers and scientists: A practical approach, pages 85–111.

Lauande, M. G. M., Teles, A. M., da Silva, L. L., Matos, C. E. F., Junior, G. B., de Paiva, A. C., de Almeida, J. D. S., da Costa Oliveira, R. M. G., Brito, H. O., Nascimento, A. G., et al. (2022). Classification of histopathological images of penile cancer using densenet and transfer learning. In VISIGRAPP (4: VISAPP), pages 976–983.

Loshchilov, I. and Hutter, F. (2017). Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101.

Machado, A. and Veloso, L. (2021). Classificação de distúrbios pulmonares em radiografias de tórax usando redes convolucionais. In Simpósio Brasileiro De Telecomunicações E Processamento De Sinais. SBC.

Melo, R. C., Raas, M. W., Palazzi, C., Neves, V. H., Malta, K. K., and Silva, T. P. (2020). Whole slide imaging and its applications to histopathological studies of liver disorders. Frontiers in medicine, 6:310.

MS (2022). Câncer de pênis: Saúde de a à z, ministério da saúde. Acesso em 13/05/2024.

Oliveira, M. V., Matos, C. E. F., Júnior, G. B., de Paiva, A. C., de Almeida, J. D. S., Costa, G., Levy, M., and Freitas, M. (2022). Ppm-unet: Uma rede neural convolucional para a segmentação de rins em imagens de tc. In Anais do XXII Simpósio Brasileiro de Computação Aplicada à Saúde, pages 268–276. SBC.

Tang, F., Wang, L., Ning, C., Xian, M., and Ding, J. (2023). Cmu-net: A strong convmixerbased medical ultrasound image segmentation network. In 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI). IEEE.

Thomas, A., Necchi, A., Muneer, A., Tobias-Machado, M., Tran, A. T. H., Van Rompuy, A.-S., Spiess, P. E., and Albersen, M. (2021). Penile cancer. Nature Reviews Disease Primers, 7(1):11.

Vahadane, A., Peng, T., Sethi, A., Albarqouni, S., Wang, L., Baust, M., Steiger, K., Schlitter, A. M., Esposito, I., and Navab, N. (2016). Structure-preserving color normalization and sparse stain separation for histological images. IEEE transactions on medical imaging, 35(8):1962–1971.