Automatic Segmentation of Skeletal Muscle Tissue in L3 CT Images Based on Random Forests and CNN Using Coarse Ground Truth Masks

• Domingos B. S. Santos UECE
• Gabriel F. L. Melo UECE
• Thelmo de Araujo UECE

DOI: https://doi.org/10.5753/sbcas.2024.1831

Resumo

Estimates of the composition of skeletal muscle tissue (SMT) and adipose tissues are important in the treatment of debilitating diseases, such as cancer, and in the control of overweight and obesity. Several studies have shown a high correlation between the percentage of SMT in computed tomography (CT) images corresponding to the cross-section at the level of the third lumbar vertebra (L3) and the percentage of this tissue in the whole body. A large number of models has been proposed to automatically segment CT images in order to estimate tissue compositions, many of them use supervised Machine Learning (ML) methods, such as neural networks, which require large amounts of labeled images, i.e., images and ground truth masks obtained from manual segmentation by human experts. These large labeled datasets are not easily available to the public, thus the present work proposes a methodology capable of performing the automatic segmentation of SMT in single-slice CT images (at L3) using only “coarse” segmentation masks as ground truth in the ML algorithms’s training phases. By “coarse segmentation” we mean a semiautomated segmentation performed by a person without specialized knowledge of human anatomy. The proposed methodology oversegments the image into superpixels, which are classified by a Random Forest (RF) model. Then, a U-Net CNN refines the classification, using as input the pixels in the superpixel segments classified as SMT by the RF. The methodology achieved 99.21% of the accuracy obtained by the same CNN trained with golden standard ground truth masks, i.e., segmentation masks manually created by a medical expert.

Referências

Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., and Süsstrunk, S. (2010). SLIC superpixels. Technical report.

Caan, B. J. et al. (2018a). Association of muscle and adiposity measured by computed tomography with survival in patients with nonmetastatic breast cancer. JAMA Oncology, 4(6):798–804.

Caan, B. J. et al. (2018b). The importance of body composition in explaining the overweight paradox in cancer-counterpoint. Cancer Research, 78(8):1906–1912.

Cheng, B. et al. (2021). Boundary IoU: Improving object-centric image segmentation evaluation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15334–15342.

Dabiri, S. et al. (2019). Muscle segmentation in axial computed tomography (CT) images at the lumbar (L3) and thoracic (T4) levels for body composition analysis. Computerized Medical Imaging and Graphics, 75:47–55.

Ding, Z. et al. (2016). Association between high visceral fat area and postoperative complications in patients with Crohn’s disease following primary surgery. Colorectal Disease, 18(2):163–172.

Liu, X. et al. (2018). Automatic organ segmentation for CT scans based on super-pixel and convolutional neural networks. Journal of Digital Imaging, 31(5):748–760.

Melo, G. F. L. (2022). Segmentação não supervisionada de tecido adiposo subcutâneo em imagens de tomografia computadorizada utilizando morfologia matemática e transformadas de distância. TCC, Universidade Estadual do Ceará, Fortaleza, Brazil.

Paris, M. T., Tandon, P., Heyland, D. K., Furberg, H., Premji, T., Low, G., and Mourtzakis, M. (2020). Automated body composition analysis of clinically acquired computed tomography scans using neural networks. Clinical Nutrition, 39(10):3049–3055.

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.

Ross, R. et al. (1992). Quantification of adipose tissue by MRI: relationship with anthropometric variables. Journal of Applied Physiology, 72(2):787–795.

Ruby, U. and Yendapalli, V. (2020). Binary cross entropy with deep learning technique for image classification. Int. J. Adv. Trends Comput. Sci. Eng, 9(10).

Schweitzer, L. et al. (2015). What is the best reference site for a single MRI slice to assess whole-body skeletal muscle and adipose tissue volumes in healthy adults? The American Journal of Clinical Nutrition, 102(1):58–65.

Schweitzer, L. et al. (2016). Estimation of skeletal muscle mass and visceral adipose tissue volume by a single magnetic resonance imaging slice in healthy elderly adults. The Journal of Nutrition, 146(10):2143–2148.