Automatic Segmentation of Deep Endometriosis in MRI Images Based on Swin-Unet

• Daniel M. Pinto UFMA
• Weslley K. R. Figueredo UFMA
• Italo F. S. da Silva UFMA
• Aristófanes C. Silva UFMA
• Anselmo C. de Paiva UFMA
• Alice C. C. B. Salomão Clínica Fonte de Imagem
• Marco A. P. de Oliveira UERJ

DOI: https://doi.org/10.5753/sbcas.2024.2715

Abstract

Deep endometriosis is the disease characterized by the presence of endometrium outside the uterine cavity, causing acute discomfort for affected individuals. Non-invasive image-based methods for assessing the degree of disease progression are effective but time-consuming for specialists. This work proposes an automatic method for segmenting endometriosis lesions in magnetic resonance images using a Swin-Unet. The method achieved a precision of 45, 6%, sensitivity of 61, 9%, dice of 47, 7% and jaccard of 36, 2%. At least one image per patient was segmented with good quality in 17 out of 18 patients used for testing.

References

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.

Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020). An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929.

Figueredo, W., Silva, I., Diniz, J., Silva, A., Paiva, A., Salomão, A., and Oliveira, M. (2023). Abordagem computacional baseada em deep learning para o diagnóstico de endometriose profunda através de imagens de ressonância magnética. In Anais do XXIII Simpósio Brasileiro de Computação Aplicada à Saúde, pages 138–149, Porto Alegre, RS, Brasil. SBC.

Kimori, Y. (2011). Mathematical morphology-based approach to the enhancement of morphological features in medical images. Journal of clinical bioinformatics, 1:1–10.

Leibetseder, A., Schoeffmann, K., Keckstein, J., and Keckstein, S. (2022). Endometriosis detection and localization in laparoscopic gynecology. Multimedia Tools and Applications, 81(5):6191–6215.

Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., and Guo, B. (2021). Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022.

Lundervold, A. S. and Lundervold, A. (2019). An overview of deep learning in medical imaging focusing on mri. Zeitschrift für Medizinische Physik, 29(2):102–127.

Manganaro, L., Fierro, F., Tomei, A., Irimia, D., Lodise, P., Sergi, M., Vinci, V., Sollazzo, P., Porpora, M., Delfini, R., et al. (2012). Feasibility of 3.0 t pelvic mr imaging in the evaluation of endometriosis. European journal of radiology, 81(6):1381–1387.

Schneider, C., Oehmke, F., Tinneberg, H.-R., and Krombach, G. (2016). Mri technique for the preoperative evaluation of deep infiltrating endometriosis: current status and protocol recommendation. Clinical Radiology, 71(3):179–194.

Wong, T.-T. (2015). Performance evaluation of classification algorithms by k-fold and leave-one-out cross validation. Pattern recognition, 48(9):2839–2846.