Segmentation of muscle tissue in computed tomography images at the level of the L3 vertebra

With the increasing routine workload on radiologists associated with the need to analyze large numbers of images, there is a need to automate part of the analysis process. Sarcopenia is a condition in which there is a loss of muscle mass. To diagnose sarcopenia, computed tomography is most often used, from the images of which the volume of muscle tissue can be assessed. The first stage of the analysis is its contouring, which is performed manually, takes a long time and is not always performed with sufficient quality affecting the accuracy of estimates and, as a result, the patient’s treatment plan. The subject of the study is the use of computer vision approaches for accurate segmentation of muscle tissue from computed tomography images for the purpose of sarcometry. The purpose of the study is to develop an approach to solving the problem of segmentation of collected and annotated images. An approach is presented that includes the stages of image pre-processing, segmentation using neural networks of the U-Net family, and post-processing. In total, 63 different configurations of the approach are considered, which differ in terms of data supplied to the input models and model architectures. The influence of the proposed method of post-processing the resulting binary masks on the segmentation accuracy is also evaluated. The approach, which includes pre-processing with table masking and anisotropic diffusion filtering, segmentation with an Inception U-Net architecture model, and post-processing based on contour analysis, achieves a Dice similarity coefficient of 0.9379 and Intersection over Union of 0.8824. Nine other configurations, the experimental results for which are reflected in the article, also demonstrated high values of these metrics (in the ranges of 0.9356–0.9374 and 0.8794–0.8822, respectively). The approach proposed in the article based on preprocessed three-channel images allows us to achieve metrics of 0.9364 and 0.8802, respectively, using the lightweight U-Net segmentation model. In accordance with the described approach, a software module was implemented in Python. The results of the study confirm the feasibility of using computer vision to assess muscle tissue parameters. The developed module can be used to reduce the routine workload on radiologists.

1. Sheberova E.V., Silanteva N.K., Agababian T.A., Potapov A.L., Nevolskikh A.A., Ivanov S.A., Kaprin A.D. Role of computed tomography in sarcopenia detection. Siberian Journal of Oncology, 2023, vol. 22, no. 3, pp. 125–133. (in Russian). https://doi.org/10.21294/1814-4861-2023-22-3-125-133

2. Agababyan T.A., Kukarskaya V.A., Silanteva N.K., Potapov A.L., Skoropad V.YU., Sheberova E.V., Dorozhkin A.D., Ivanov S.A., Kaprin A.D. CT sarcometry in the prediction of postoperative complications in patients with gastric cancer: retrospective cohort study. Journal of Modern Oncology, 2023, vol. 25, no. 3, pp. 284–288. (in Russian). https://doi.org/10.26442/18151434.2023.3.202260

3. Tagliafico A.S., Bignotti B., Torri L., Rossi F. Sarcopenia: how to measure, when and why // La radiologia medica. 2022. V. 127. N 3. P. 228–237. https://doi.org/10.1007/s11547-022-01450-3

4. Shen W., Punyanitya M., Wang Z., Gallagher D., St-Onge M.P., Albu J., Heymsfield S.B., Heshka S. Total body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image // Journal of Applied Physiology. 2004. V. 97. N 6. P. 2333–2338. https://doi.org/10.1152/japplphysiol.00744.2004

5. Van den Broeck J., Sealy M.J., Brussaard C., Kooijman J., Jager-Wittenaar H., Scafoglieri A. The correlation of muscle quantity and quality between all vertebra levels and level L3, measured with CT: An exploratory study // Frontiers in Nutrition. 2023. V. 10. P. 1148809. https://doi.org/10.3389/fnut.2023.1148809

6. Vedire Y., Nitsche L., Tiadjeri M., McCutcheon V., Hall J., Barbi J., Yendamuri S., Ray A.D. Skeletal muscle index is associated with long term outcomes after lobectomy for non-small cell lung cancer // BMC Cancer. 2023. V. 23. N 1. P. 778. https://doi.org/10.1186/s12885-023-11210-9

7. Smorchkova A.K., Petraikin A.V., Semenov D.S., Sharova D.E. Sarcopenia: Modern approaches to solving diagnosis problems. Digital Diagnostics, 2022, vol. 3, no. 3, pp. 196–211. https://doi.org/10.17816/DD110721

8. Burns J.E., Yao J., Chalhoub D., Chen J.J., Summers R.M. A machine learning algorithm to estimate sarcopenia on abdominal CT // Academic Radiology. 2020. V. 27. N 3. P. 311–320. https://doi.org/10.1016/j.acra.2019.03.011

9. Graffy P.M., Liu J., Pickhardt P.J., Burns J.E., Yao J., Summers R.M. Deep learning-based muscle segmentation and quantification at abdominal CT: application to a longitudinal adult screening cohort for sarcopenia assessment // British Journal of Radiology. 2019. V. 92. N 1100. P. 20190327. https://doi.org/10.1259/bjr.20190327

10. Blanc-Durand P., Schiratti J.B., Schutte K., Jehanno P., Herent P., Pigneur F., Lucidarme O., Benaceur Y., Sadate A., Luciani A., Ernst O., Rouchaud A., Creze M., Dallongeville A., Banaste N., Cadi M., Bousaid I., Lassau N., Jegou S. Abdominal musculature segmentation and surface prediction from CT using deep learning for sarcopenia assessment // Diagnostic and Interventional Imaging. 2020. V. 101. N 12. P. 789–794. https://doi.org/10.1016/j.diii.2020.04.011

11. Ackermans L.L.G.C., Volmer L., Wee L., Brecheisen R., Sánchez-González P., Seiffert A.P., Gómez E.J., Dekker A., Ten Bosch J.A., Olde Damink S.M.W., Blokhuis T.J. Deep learning automated segmentation for muscle and adipose tissue from abdominal computed tomography in polytrauma patients // Sensors. 2021. V. 21. N 6. P. 2083. https://doi.org/10.3390/s21062083

12. Kreher R., Hinnerichs M., Preim B., Saalfeld S., Surov A. Deep-learning-based Segmentation of Skeletal Muscle Mass in Routine Abdominal CT Scans // In Vivo. 2022. V. 36. N 4. P. 1807–1811. https://doi.org/10.21873/invivo.12896

13. Song G., Zhou J., Wang K., Yao D., Chen S., Shi Y. Segmentation of multi-regional skeletal muscle in abdominal CT image for cirrhotic sarcopenia diagnosis // Frontiers in Neuroscience. 2023. V. 17. P. 1203823. https://doi.org/10.3389/fnins.2023.1203823

14. Dabiri S., Popuri K., Cespedes Feliciano E.M., Caan B.J., Baracos V.E., Beg M.F. 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. 2019. V. 75. P. 47–55. https://doi.org/10.1016/j.compmedimag.2019.04.007

15. Gu S., Wang L., Han R., Liu X., Wang Y., Chen T., Zheng Z. Detection of sarcopenia using deep learning-based artificial intelligence body part measure system (AIBMS) // Frontiers in Physiology. 2023. V. 14. P. 1092352. https://doi.org/10.3389/fphys.2023.1092352

16. Islam S., Kanavati F., Arain Z., Da Costa O.F., Crum W., Aboagye E.O., Rockall A.G. Fully automated deep-learning section-based muscle segmentation from CT images for sarcopenia assessment // Clinical Radiology. 2022. V. 77. N 5. P. e363–e371. https://doi.org/10.1016/j.crad.2022.01.036

17. Takahashi N., Sugimoto M., Psutka S.P., Chen B., Moynagh M.R., Carter R.E. Validation study of a new semi-automated software program for CT body composition analysis // Abdominal Radiology. 2017. V. 42. N 9. P. 2369–2375. https://doi.org/10.1007/s00261-017-1123-6

18. Kaur R., Juneja M., Mandal A.K. A comprehensive review of denoising techniques for abdominal CT images // Multimedia Tools and Applications. 2018. V. 77. N 17. P. 22735–22770. https://doi.org/10.1007/s11042-017-5500-5

19. Masenko V.L., Kokov A.N., Grigoreva I.I., Krivoshapova K.E. Radiology methods of the sarcopenia diagnosis. Research and Practical Medicine Journal, 2019, vol. 6, no. 4, pp. 127–137. (in Russian). https://doi.org/10.17709/2409-2231-2019-6-4-13

20. Lee J.S., Kim Y.S., Kim E.Y., Jin W. Prognostic significance of CT-determined sarcopenia in patients with advanced gastric cancer // PLoS ONE. 2018. V. 13. N 8. P. e0202700. https://doi.org/10.1371/journal.pone.0202700

21.