Weighted Deep Learning Ensemble for Breast Lesion Segmentation in Ultrasound Images

• Neilson P. Ribeiro IFMA / UFMA
• Celso L. S. Soares Filho UFMA
• Felipe R. S. Teles UFMA
• Marcos R. A. Amorim UFMA
• João O. B. Diniz IFMA / UFMA
• Anselmo C. de Paiva UFMA
• Aristófanes C. Silva UFMA
• Antonio O. C. Filho UFPI

DOI: https://doi.org/10.5753/sbcas.2026.21356

Abstract

Breast cancer is one of the leading causes of mortality among women worldwide, making early diagnosis essential to improve survival rates. In this context, automatic lesion segmentation in ultrasound images remains a challenging task due to speckle noise, low contrast, and high morphological variability. This work proposes a weighted ensemble approach combining U-Net++, DeepLabV3+, and Swin-UNet architectures, leveraging complementary features to enhance prediction robustness. Evaluated on the BUS-BRA (INCA) dataset, the proposed method achieved an IoU of 85.62% and a Dice score of 93.17%, outperforming previously reported results in the literature. The findings demonstrate the potential of the proposed ensemble strategy as a clinical decision-support tool.

References

Abraham, N. and Khan, N. M. (2019). A novel focal tversky loss function with improved attention u-net for lesion segmentation. In 2019 IEEE 16th international symposium on biomedical imaging (ISBI 2019), pages 683–687. IEEE.

Cao, H., Wang, Y., Chen, J., Jiang, D., Zhang, X., Tian, Q., and Wang, M. (2022). Swin-unet: Unet-like pure transformer for medical image segmentation. In European conference on computer vision, pages 205–218. Springer.

Chen, L.-C., Zhu, Y., Papandreou, G., Schroff, F., and Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. In Proceedings of the European conference on computer vision (ECCV), pages 801–818.

Diniz, J. O., Dias Jr, D. A., da Cruz, L. B., Marques, R. C., Gomes Jr, D. L., Cortês, O. A., de Carvalho Filho, A. O., and Quintanilha, D. B. (2024). Efficientensemble: Diagnóstico de câncer de mama em imagens de ultrassom utilizando processamento de imagens e ensemble de efficientnets. In Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS), pages 202–213. SBC.

Gómez-Flores, W., Gregorio-Calas, M. J., and Coelho de Albuquerque Pereira, W. (2024). Bus-bra: a breast ultrasound dataset for assessing computer-aided diagnosis systems. Medical Physics, 51(4):3110–3123.

Gonçalves, J. V., de Souza, D. V., dos Santos, C. I., do Nascimento, C. E., da Cruz, L. B., Junior, D. A., and Diniz, J. O. (2024). D. iagnóstica: Ferramenta cadx para diagnóstico de doenças pulmonares em imagens radiológicas. In Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS), pages 214–225. SBC.

Gonzalez, R. and Woods, R. (2000). Processamento de imagens digitais. Edgard Blucher.

Guo, Y., Qiang, Y., Chen, Q., Li, Q., and Sun, J. (2025). Msra-net: A multi-scale and region-aware network for breast cancer ultrasound image segmentation. Digital Signal Processing, page 105534.

Kim, J., Harper, A., McCormack, V., Sung, H., Houssami, N., Morgan, E., Mutebi, M., Garvey, G., Soerjomataram, I., and Fidler-Benaoudia, M. M. (2025). Global patterns and trends in breast cancer incidence and mortality across 185 countries. Nature Medicine, pages 1–9.

Litjens, G., Kooi, T., Bejnordi, B. E., Setio, A. A. A., Ciompi, F., Ghafoorian, M., Van Der Laak, J. A., Van Ginneken, B., and Sánchez, C. I. (2017). A survey on deep learning in medical image analysis. Medical image analysis, 42:60–88.

Qiu, W., Hamburg, E., Zhou, Y., and Salehani, Y. E. (2025). Lightweight u-net for breast ultrasound image segmentation.

Ronneberger, O., Fischer, P., and Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention, pages 234–241. Springer.

Sood, R., Rositch, A. F., Shakoor, D., Ambinder, E., Pool, K.-L., Pollack, E., Mollura, D. J., Mullen, L. A., and Harvey, S. C. (2019). Ultrasound for breast cancer detection globally: a systematic review and meta-analysis. Journal of global oncology.

Teles, F. R., Ribeiro, N. P., da Cruz, L. B., Júnior, G. B., de Paiva, A. C., Diniz, J. O., and Cortes, O. A. (2025). Atlas-guided u-net++ with efficientnetb5 for automatic pancreas segmentation in abdominal ct scans. In Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS), pages 104–115. SBC.

Wei, W., Wu, J., and Shao, G. (2025). Research on breast tumor segmentation based on the mamba architecture. Frontiers in Oncology, 15:1672274.

Zhang, C., Wang, L., Wei, G., Kong, Z., and Qiu, M. (2024). A dual-branch and dual attention transformer and cnn hybrid network for ultrasound image segmentation. Frontiers in Physiology, 15:1432987.

Zhou, Z., Rahman Siddiquee, M. M., Tajbakhsh, N., and Liang, J. (2018). Unet++: A nested u-net architecture for medical image segmentation. In International workshop on deep learning in medical image analysis, pages 3–11. Springer.