DOI QR코드

DOI QR Code

Development and Validation of AI Image Segmentation Model for CT Image-Based Sarcopenia Diagnosis

CT 영상 기반 근감소증 진단을 위한 AI 영상분할 모델 개발 및 검증

  • 이충섭 (원광대학교 의료융합연구센터 ) ;
  • 임동욱 (원광대학교 의료융합연구센터 ) ;
  • 노시형 (원광대학교 의료융합연구센터) ;
  • 김태훈 (원광대학교 의료융합연구센터) ;
  • 고유선 (울산대학교 의학과 연구조) ;
  • 김경원 (서울아산병원 영상의학과 ) ;
  • 정창원 (원광대학교병원 의생명연구원 중점연구센터 )
  • Received : 2022.07.29
  • Accepted : 2022.10.26
  • Published : 2023.03.31

Abstract

Sarcopenia is not well known enough to be classified as a disease in 2021 in Korea, but it is recognized as a social problem in developed countries that have entered an aging society. The diagnosis of sarcopenia follows the international standard guidelines presented by the European Working Group for Sarcopenia in Older People (EWGSOP) and the d Asian Working Group for Sarcopenia (AWGS). Recently, it is recommended to evaluate muscle function by using physical performance evaluation, walking speed measurement, and standing test in addition to absolute muscle mass as a diagnostic method. As a representative method for measuring muscle mass, the body composition analysis method using DEXA has been formally implemented in clinical practice. In addition, various studies for measuring muscle mass using abdominal images of MRI or CT are being actively conducted. In this paper, we develop an AI image segmentation model based on abdominal images of CT with a relatively short imaging time for the diagnosis of sarcopenia and describe the multicenter validation. We developed an artificial intelligence model using U-Net that can automatically segment muscle, subcutaneous fat, and visceral fat by selecting the L3 region from the CT image. Also, to evaluate the performance of the model, internal verification was performed by calculating the intersection over union (IOU) of the partitioned area, and the results of external verification using data from other hospitals are shown. Based on the verification results, we tried to review and supplement the problems and solutions.

근감소증은 국내는 2021년 질병으로 분류되었을 만큼 잘 알려져 있지 않지만 고령화사회에 진입한 선진국에서는 사회적 문제로 인식하고 있다. 근감소증 진단은 유럽노인근감소증 진단그룹(EWGSOP)과 아시아근감소증진단그룹(AWGS)에서 제시하는 국제표준지침을 따른다. 최근 진단방법으로 절대적 근육량 이외에 신체수행평가로 보행속도 측정과 일어서기 검사 등을 통하여 근육 기능을 함께 측정할 것을 권고하고 있다. 근육량을 측정하기 위한 대표적인 방법으로 DEXA를 이용한 체성분 분석 방법이 임상에서 정식으로 실시하고 있다. 또한 MRI 또는 CT의 복부 영상을 이용하여 근육량을 측정하는 다양한 연구가 활발하게 진행되고 있다. 따라서 본 논문에서는 근감소증 진단을 위해서 비교적 짧은 촬영시간을 갖는 CT의 복부영상기반으로 AI 영상 분할 모델을 개발하고 다기관 검증한 내용을 기술한다. 우리는 CT 영상 중에 요추의 L3 영역을 분류하여 피하지방, 내장지방, 근육을 자동으로 분할할 수 있는 인공지능 모델을 U-Net 모델을 사용하여 개발하였다. 또한 모델의 성능평가를 위해서 분할영역의 IOU(Intersection over Union)를 계산하여 내부검증을 진행했으며, 타 병원의 데이터를 활용하여 동일한 IOU 방법으로 외부검증을 진행한 결과를 보인다. 검증 결과를 토대로 문제점과 해결방안에 대해서 검증하고 보완하고자 했다.

Keywords

Acknowledgement

본 연구는 보건복지부의 재원으로 한국보건산업진흥원의 보건의료기술 연구개발사업(HI18C1216), 그리고 한국 연구재단(NRF-2021R1A5A 8029876)(NRF-2020R1I1A1A01074256)의 지원으로 수행함.

References

  1. J.-Y. Lim, "Consensus update on definition and diagnosis for sarcopenia," Geriatric Rehabilitation, Vol.10, No.2, pp.39-45, 2020.
  2. G. Grimby and B. Saltin, "The ageing muscle," Clin Physiol, Vol.3, No.3, pp.209-18, 1983. doi: 10.1111/j.1475-097x.1983.tb00704 .x. PMID: 6347501.
  3. A. J. Cruz-Jentoft et al., "Sarcopenia: European consensus on definition and diagnosis: Report of the European Working Group on Sarcopenia in Older People," Age Ageing, Vol.39, No.4, pp.412-23, 2010. doi: 10.1093/ageing/afq034. Epub 2010 Apr 13. PMID: 20392703; PMCID: PMC2886201.
  4. O. Wallengren et al., "Comparison of the 2010 and 2019 diagnostic criteria for sarcopenia by the European Working Group on Sarcopenia in Older People (EWGSOP) in two cohorts of Swedish older adults," BMC Geriatr, Vol.21, No.1, pp.600, 2021. https://doi.org/10.1186/s12877-021-02533-y.
  5. L. K. Chen et al., "Sarcopenia in Asia: Consensus report of the Asian Working Group for Sarcopenia," Journal of the American Medical Directors Association, Vol.15, No.2, pp.95-101, 2014. https://doi.org/10.1016/j.jamda.2013.11.025
  6. L. K. Chen et al., "Asian Working Group for Sarcopenia: 2019 consensus update on sarcopenia diagnosis and treatment," Journal of the American Medical Directors Association, Vol.21, No.3, pp.300-307, e2, 2020. https://doi.org/10.1016/j.jamda.2019.12.012
  7. C.-S. Lee et al., "Implementation of medical image labeling web application for machine learning," Proceedings of the Korea Information Processing Society Conference, pp.602-605. 2021, doi:10.3745/PKIPS.Y2021M11A.602.
  8. R. B. Parikh, Z. Obermeyer, and A. S. Navathe, "Regulation of predictive analytics in medicine," Science, Vol.363, No. 6429, pp.810-812, 2019. doi:10.1126/science.aaw0029.
  9. C. L Ramspek, K. J. Jager, F. W. Dekker, C. Zoccali, and M. van Diepen, "External validation of prognostic models: What, why, how, when and where," Clinical Kidney Journal, Vol.14, Iss.1, pp.49-58, 2021. https://doi.org/10.1093/ckj/sfaa188.
  10. S. E. Bleeker et al., "External validation is necessary in prediction research: A clinical example," Journal of clinical epidemiology, Vol.56, No.9, pp.826-832, 2003. https://doi.org/10.1016/S0895-4356(03)00207-5
  11. L. L. G. C. Ackermans et al., "Screening, diagnosis and monitoring of sarcopenia: When to use which tool?" Clinical Nutrition ESPEN, Vol.48, pp.36-44, 2022. https://doi.org/10.1016/j.clnesp.2022.01.027
  12. V. E. Baracos, V. C. Mazurak, and A. S. Bhullar, "Cancer cachexia is defined by an ongoing loss of skeletal muscle mass," Annals of Palliative Medicine, Vol.8, No.1, pp.3-12, 2018. https://doi.org/10.21037/apm.2018.12.01
  13. M. T. Imboden, A. M. Swartz, H. W. Finch, M. P. Harber, and L. A. Kaminsky, "Reference standards for lean mass measures using GE dual energy x-ray absorptiometry in Caucasian adults," PLoS One, Vol.12, No.4, Article e0176161, 2017.
  14. K. Engelke, O. Museyko, L. Wang, and J.-D. Laredo, "Quantitative analysis of skeletal muscle by computed tomography imaging-state of the art," Journal of orthopaedic translation, Vol.15, pp.91-103, 2018. https://doi.org/10.1016/j.jot.2018.10.004
  15. S. Bamba et al., "Sarcopenia is a predictive factor for intestinal resection in admitted patients with Crohn's disease," PLoS One, Vol.12, Vol.6, e0180036, 2017.
  16. R. Hemke, C. G. Buckless, A. Tsao, B. Wang, and M. Torriani, "Deep learning for automated segmentation of pelvic muscles, fat, and bone from CT studies for body composition assessment," Skeletal Radiology, Vol.49, No.3, pp.387-395, 2020. https://doi.org/10.1007/s00256-019-03289-8
  17. L. Boeri et al., "Hounsfield unit attenuation value can differentiate pyonephrosis from hydronephrosis and predict septic complications in patients with obstructive uropathy," Scientific Reports, Vol.10, No.1, pp.18546, 2020. https://doi.org/10.1038/s41598-020-75672-8.
  18. J. Schmidhuber, "Deep learning in neural networks: An overview," Neural Networks, Vol.61, pp.85-117, 2015. doi: 10.1016/j.neunet.2014.09.003. Epub 2014 Oct. 13. PMID:25462637.
  19. O. Ronneberger, P. Fischer, and T. Brox, "U-Net: Convolutional networks for biomedical image segmentation," arXiv preprint arXiv:1505.04597, 2015. 5.
  20. S.-H. Noh et al., "Development of image segmentation model for sarcopenia diagnosis and its application," Proceedings of the Korea Information Processing Society Conference, pp.577-79, 2021, doi:10.3745/PKIPS.Y2021M11A.577.
  21. C. H. Sudre, W. Li, T. Vercauteren, S. Ourselin, and M. J. Cardoso, "Generalized dice overlap as a deep learning loss function for highly unbalanced segmentations," in Proceeding of Deep Learning in Medical Image Analysis and Multimodal Laerning for Clinical Decision Support, pp.240-248, 2017.
  22. Z. Zhang and M. Sabuncu, "Generalized cross entropy loss for training deep neural networks with noisy labels," arXiv preprint arXiv:1805.07836, 2018.
  23. J. M. Johnson and T. M. Khoshgoftaar, "Survey on deep learning with class imbalance," Journal of Big Data, Vol.6, No.1, pp.1-5, 2019. https://doi.org/10.1186/s40537-019-0192-5.
  24. M. Buda, A. Maki, and M. A. Mazurowski, "A systematic study of the class imbalance problem in convolutional neural networks," CoRR abs/1710.05381 arXiv:1710.05381, 2017. http://arxiv.org/abs/1710.05381.
  25. H. J. Park et al., "Development and validation of a deep learning system for segmentation of abdominal muscle and fat on computed tomography," Korea Journal of Radiology, Vol.21, No.1, pp.88-100, 2020. http://doi.org/10.3348/kjr.2019.0470.