Segmentação de coração em tomografias computadorizadas utilizando atlas probabilístico e redes neurais convolucionais

• João O. B. Diniz UFMA / IFMA
• Jonnison L. Ferreira UFMA
• Giovanni L. F. da Silva UFMA / UNDB
• Darlan B. P. Quintanilha UFMA
• Aristófanes C. Silva UFMA
• Anselmo Paiva UFMA

DOI: https://doi.org/10.5753/sbcas.2021.16055

Resumo

Órgãos em risco (OARs) são tecidos saudáveis ao redor do câncer que devem ser preservados na radioterapia (RT). O coração é um dos OARs fundamentais, assim, softwares computacionais foram desenvolvidos para auxiliar os especialistas na segmentação. Neste trabalho, propõe-se um método automático para segmentação a partir da tomografia computadorizada. O método consiste em 3 etapas: (1) aquisição de banco de dados público e diversificado; (2) padronização de volume usando registro e correspondência de histograma; e (3) segmentação do coração usando atlas e U-Net com blocos residuais (ResU-Net). Assim, alcançou-se 92,53% de Dice e 84,73% de Jaccard. Com a inovação e os resultados, mostra-se que o método proposto é promissor.

Referências

Button, M. and Staffurth, J. (2010). Clinical application of image-guided radiotherapy in bladder and prostate cancer. Clinical Oncology, 22(8):698–706.

Chen, P., Xu, C., Li, X., Ma, Y., and Sun, F. (2019). Two-stage Network for OAR segmentation.

da Cruz, L. B., Araújo, J. D. L., Ferreira, J. L., Diniz, J. O. B., Silva, A. C., de Almeida, J. D. S., de Paiva, A. C., and Gattass, M. (2020). Kidney segmentation from computed tomography images using deep neural network. Computers in Biology and Medicine, page 103906.

Diniz, J. O. B., Diniz, P. H. B., Valente, T. L. A., Silva, A. C., and Paiva, A. C. (2019). Spinal cord detection in planning ct for radiotherapy through adaptive template matching, imslic and convolutional neural networks. Computer methods and programs in biomedicine, 170:53–67.

Diniz, J. O. B., Ferreira, J. L., Diniz, P. H. B., Silva, A. C., and de Paiva, A. C. (2020). Esophagus segmentation from planning ct images using an atlas-based deep learning approach. Computer Methods and Programs in Biomedicine, page 105685.

Diniz, P. H. B., Valente, T. L. A., Diniz, J. O. B., Silva, A. C., Gattass, M., Ventura, N., Muniz, B. C., and Gasparetto, E. L. (2018). Detection of white matter lesion regions in mri using slic0 and convolutional neural network. Computer methods and programs in biomedicine, 167:49–63.

Dong, X., Lei, Y., Wang, T., Thomas, M., Tang, L., Curran, W. J., Liu, T., and Yang, X. (2019). Automatic multiorgan segmentation in thorax ct images using u-net-gan. Medical physics, 46(5):2157–2168.

Evans, E. and Staffurth, J. (2018). Principles of cancer treatment by radiotherapy. Surgery-Oxford International Edition, 36(3):111–116.

Feng, X., Bernard, M. E., Hunter, T., and Chen, Q. (2020). Improving accuracy and robustness of deep convolutional neural network based thoracic oar segmentation. Physics in Medicine & Biology, 65(7):07NT01.

Gómez-Millán, J., Lara, M. F., Generoso, R. C., Perez-Rozos, A., Lupiáñez-Pérez, Y., and Carmona, J. A. M. (2015). Advances in the treatment of prostate cancer with radiotherapy. Critical reviews in oncology/hematology, 95(2):144–153.

Gonzalez, R. C. and Woods, R. E. (2010). Digital Image Processing. Pearson Prentice Hall.

La Macchia, M., Fellin, F., Amichetti, M., Cianchetti, M., Gianolini, S., Paola, V., Lomax, A. J., and Widesott, L. (2012). Systematic evaluation of three different commercial software solutions for automatic segmentation for adaptive therapy in head-and-neck, prostate and pleural cancer. Radiation Oncology, 7(1):160.

Paiva, A. M., Diniz, J. O., Silva, A., and Paiva, A. (2019). Segmentação de vértebras e diagnóstico de fraturas em imagens de ressonância magnética utilizando u-net 3d e deep belief network. In Anais do XIX Simpósio Brasileiro de Computação Aplicada à Saúde, pages 106–117, Porto Alegre, RS, Brasil. SBC.

Piroth, M. D., Baumann, R., Budach, W., Dunst, J., Feyer, P., Fietkau, R., Haase, W., Harms, W., Hehr, T., Krug, D., et al. (2019). Heart toxicity from breast cancer radiotherapy. Strahlentherapie und Onkologie, 195(1):1–12.

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.

St Germain, D., Denicoff, A., Torres, A., Kelaghan, J., McCaskill-Stevens, W., Mishkin, G., O’Mara, A., and Minasian, L. M. (2020). Reporting of health-related quality of life endpoints in national cancer institute–supported cancer treatment trials.

Sudre, C. H., Li, W., Vercauteren, T., Ourselin, S., and Cardoso, M. J. (2017). Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations. In Deep learning in medical image analysis and multimodal learning for clinical decision support, pages 240–248. Springer.

Trullo, R., Petitjean, C., Nie, D., Shen, D., and Ruan, S. (2017). Joint segmentation of multiple thoracic organs in ct images with two collaborative deep architectures. In Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, pages 21–29. Springer.

Tsang, Y. M. and Hoskin, P. (2017). The impact of bladder preparation protocols on post treatment toxicity in radiotherapy for localised prostate cancer patients. Technical Innovations and Patient Support in Radiation Oncology, 3:37–40.

Vesal, S., Ravikumar, N., and Maier, A. (2019). A 2d dilated residual u-net for multiorgan segmentation in thoracic ct. arXiv preprint arXiv:1905.07710.

Yoo, T. S. (2004). Insight into images: principles and practice for segmentation, registration, and image analysis. AK Peters/CRC Press.

Zhang, J., Yang, Y., Shao, K., Bai, X., Fang, M., Shan, G., and Chen, M. (2020). A unet- based research on the multi-output convolution neural network's ability of decreasing mis-identication: Automatic segmentation of organs at risk in thorax.