대한영상의학회지 (Journal of the Korean Society of Radiology)

  • 제83권6호
  • /
  • Pages.1298-1311
  • /
  • 2022
  • /
  • 1738-2637(pISSN)
  • /
  • 2288-2928(eISSN)
대한영상의학회 (The Korean Society of Radiology)
권호 표지
DOI QR코드

DOI QR Code

준지도학습 방법을 이용한 흉부 X선 사진에서 척추측만증의 진단

Diagnosis of Scoliosis Using Chest Radiographs with a Semi-Supervised Generative Adversarial Network

  • 이우진 (한양대학교병원 영상의학과) ;
  • 신기원 (울산대학교 의과대학 서울아산병원 아산융합의학원 의공학과) ;
  • 이준수 (서울대학교 산업공학과) ;
  • 유승진 (한양대학교병원 영상의학과) ;
  • 윤민아 (울산대학교 의과대학 서울아산병원 영상의학과) ;
  • 최요원 (한양대학교병원 영상의학과) ;
  • 홍길선 (울산대학교 의과대학 서울아산병원 영상의학과) ;
  • 김남국 (울산대학교 의과대학 서울아산병원 아산융합의학원 의공학과) ;
  • 백상현 (한양대학교병원 영상의학과)
  • Woojin Lee (Department of Radiology, Hanyang University Hospital) ;
  • Keewon Shin (Department of Bioengineering, Asan Medical Institute of Convergence Science and Technology, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Junsoo Lee (Department of Industrial Engineering, Seoul National University) ;
  • Seung-Jin Yoo (Department of Radiology, Hanyang University Hospital) ;
  • Min A Yoon (Department of Radiology, University of Ulsan College of Medicine, Asan Medical Center) ;
  • Yo Won Choi (Department of Radiology, Hanyang University Hospital) ;
  • Gil-Sun Hong (Department of Radiology, University of Ulsan College of Medicine, Asan Medical Center) ;
  • Namkug Kim (Department of Bioengineering, Asan Medical Institute of Convergence Science and Technology, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Sanghyun Paik (Department of Radiology, Hanyang University Hospital)
  • 투고 : 2021.08.27
  • 심사 : 2021.11.08
  • 발행 : 2022.11.01
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

목적 흉부 X선 사진에서 척추측만증을 조기진단 할 수 있는 딥러닝 기반의 스크리닝 소프트웨어를 준지도학습(semi-supervised generative adversarial network; 이하 GAN) 방법을 이용하여 개발하고자 하였다. 대상과 방법 두 곳의 상급종합병원에서 촬영된 흉부 X선 사진에서 척추측만증을 조기진단할 수 있는 스크리닝 소프트웨어를 개발하기 위하여 GAN 방법이 이용되었다. GAN의 훈련과정에서 경증에서 중증의 척추측만증을 보이는 흉부 X선 사진들을 사용하였으며 upstream task에서 척추측만증의 특징을 학습하고, downstream task에서 정상과 척추측만증을 분류하도록 훈련하였다. 결과 수신자 조작 특성 곡선의 곡선하면적(area under the receiver operating characteristic curve), 음성예측도, 양성예측도, 민감도 및 특이도는 각각 0.856, 0.950, 0.579, 0.985, 0.285이었다. 결론 우리가 GAN 방법을 이용하여 개발한 딥러닝 기반의 스크리닝 소프트웨어는 청소년의 흉부 X선에서 척추측만증을 진단하는데 있어서 높은 음성예측도와 민감도를 보였다. 이 소프트웨어가 건강검진을 목적으로 촬영한 청소년의 흉부 X선 사진에 진단 스크리닝 도구로써 이용된다면 영상의학과 의사의 부담을 덜어주며, 척추측만증의 조기진단에 기여할 것으로 생각된다.

Purpose To develop and validate a deep learning-based screening tool for the early diagnosis of scoliosis using chest radiographs with a semi-supervised generative adversarial network (GAN). Materials and Methods Using a semi-supervised learning framework with a GAN, a screening tool for diagnosing scoliosis was developed and validated through the chest PA radiographs of patients at two different tertiary hospitals. Our proposed method used training GAN with mild to severe scoliosis only in a semi-supervised manner, as an upstream task to learn scoliosis representations and a downstream task to perform simple classification for differentiating between normal and scoliosis states sensitively. Results The area under the receiver operating characteristic curve, negative predictive value (NPV), positive predictive value, sensitivity, and specificity were 0.856, 0.950, 0.579, 0.985, and 0.285, respectively. Conclusion Our deep learning-based artificial intelligence software in a semi-supervised manner achieved excellent performance in diagnosing scoliosis using the chest PA radiographs of young individuals; thus, it could be used as a screening tool with high NPV and sensitivity and reduce the burden on radiologists for diagnosing scoliosis through health screening chest radiographs.

키워드

과제정보

We would like to thank Editage (www.editage.co.kr) for English language editing.

참고문헌

  1. Kim H, Kim HS, Moon ES, Yoon CS, Chung TS, Song HT, et al. Scoliosis imaging: what radiologists should know. Radiographics 2010;30:1823-1842
  2. Roach JW. Adolescent idiopathic scoliosis. Orthop Clin North Am 1999;30:353-365
  3. Baert AL. Spinal imaging: diagnostic imaging of the spine and spinal cord. Berlin: Springer Science & Business Media 2007
  4. Silva FE, Lenke LG. Adolescent idiopathic scoliosis. In Errico TJ, Lonner BS, Moulton AW, eds. Surgical management of spinal deformities. Philadelphia: Saunders Elsevier, 2009:97-118
  5. Weinstein SL, Zavala DC, Ponseti IV. Idiopathic scoliosis: long-term follow-up and prognosis in untreated patients. J Bone Joint Surg Am 1981;63:702-712
  6. Malfair D, Flemming AK, Dvorak MF, Munk PL, Vertinsky AT, Heran MK, et al. Radiographic evaluation of scoliosis: review. AJR Am J Roentgenol 2010;194:S8-S22
  7. Pruijs JE, Hageman MA, Keessen W, van der Meer R, van Wieringen JC. Variation in Cobb angle measurements in scoliosis. Skeletal Radiol 1994;23:517-520
  8. Morrissy RT, Goldsmith GS, Hall EC, Kehl D, Cowie GH. Measurement of the Cobb angle on radiographs of patients who have scoliosis. Evaluation of intrinsic error. J Bone Joint Surg Am 1990;72:320-327
  9. Galbusera F, Niemeyer F, Wilke HJ, Bassani T, Casaroli G, Anania C, et al. Fully automated radiological analysis of spinal disorders and deformities: a deep learning approach. Eur Spine J 2019;28:951-960
  10. Alharbi RH, Alshaye MB, Alkanhal MM, Alharbi NM, Alzahrani MA, Alrehaili OA.Deep learning based algorithm for automatic scoliosis angle measurement. Proceedings of 2020 3rd International Conference on Computer Applications & Information Security (ICCAIS); 2020 Mar 19-21; Riyadh, Saudi Arabia: IEEE; 2020:1-5
  11. Tan Z, Yang K, Sun Y, Wu B, Tao H, Hu Y, et al. An automatic scoliosis diagnosis and measurement system based on deep learning. Proceedings of the ROBIO 2018: IEEE International Conference on Robotics and Biomimetics; 2018 Dec 12-15; Kuala Lumpur, Malaysia: IEEE; 2018:439-443
  12. Reza AM. Realization of the contrast limited adaptive histogram equalization (CLAHE) for real-time image enhancement. J VLSI Signal Process Syst Signal Image Video Technol 2004;38:35-44
  13. Karras T, Aittala M, Hellsten J, Laine S, Lehtinen J, Aila T. Training generative adversarial networks with limited data. arXiv 2006.06676 [Preprint]. 2020 [cited 2020 October 7]. Available at. https://arxiv.org/abs/2006.06676
  14. Gardner MW, Dorling SR. Artificial neural networks (the multilayer perceptron)-a review of applications in the atmospheric sciences. Atmos Environ 1998;32:2627-2636
  15. Liu Z, Luo P, Wang X, Tang X. Deep learning face attributes in the wild. In Proceedings of the IEEE International Conference on Computer Vision; 2015 December 13-16; Santiago, Chile: IEEE; 2015:3730-3738
  16. Karras T, Laine S, Aila T. A style-based generator architecture for generative adversarial networks. Proceedings of the Conference on Computer Vision and Pattern Recognition (CVPR); 2019 Jun 16-20; Long Beach, CA, USA: IEEE; 2019:4401-4410
  17. Heusel M, Ramsauer H, Unterthiner T, Nessler B, Hochreiter S. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems. Proceedings of the 31st Conference on Neural Information Processing Systems (NIPS 2017); 2017 Dec 4-9; Long Beach, CA, USA: NIPS; 2017:30
  18. Kingma DP, Ba J. Adam: a method for stochastic optimization. ArXiv 1412.6980 [Preprint]. 2014 [cited 2019 May 1]. Available at. https://arxiv.org/abs/1412.6980
  19. He K, Zhang X, Ren S, Sun J. Identity mappings in deep residual networks. ECCV 2016. Lecture notes in computer science, vol 9908. Cham: Springer 2016
  20. Tu Y, Wang N, Tong F, Chen H. Automatic measurement algorithm of scoliosis Cobb angle based on deep learning. J Phys Conf Ser 2019;1187:042100
  21. Liu T, Yang Y, Wang Y, Sun M, Fan W, Wu C, et al. Spinal curve assessment of idiopathic scoliosis with a small dataset via a multi-scale keypoint estimation approach. Adjunct Proceedings of the 2020 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2020 ACM International Symposium on Wearable Computers; 2020 Sep 12-17; Virtual Event, Mexico: Association for Computing Machinery; 2020:665-670
  22. Kokabu T, Kanai S, Kawakami N, Uno K, Kotani T, Suzuki T, et al. An algorithm for using deep learning convolutional neural networks with three dimensional depth sensor imaging in scoliosis detection. Spine J 2021;21:980-987
  23. Yang J, Zhang K, Fan H, Huang Z, Xiang Y, Yang J, et al. Development and validation of deep learning algorithms for scoliosis screening using back images. Commun Biol 2019;2:390
  24. Zenati H, Foo CS, Lecouat B, Manek G, Chandrasekhar VR. Efficient gan-based anomaly detection. ArXiv 1802.06222 [Preprint]. 2018 [cited 2019 May 1]. Available at. https://arxiv.org/abs/1802.06222
  25. Schlegl T, Seebock P, Waldstein SM, Langs G, Schmidt-Erfurth U. f-AnoGAN: fast unsupervised anomaly detection with generative adversarial networks. Med Image Anal 2019;54:30-44
  26. Oh CH, Kim CG, Lee MS, Yoon SH, Park HC, Park CO. Usefulness of chest radiographs for scoliosis screening: a comparison with thoraco-lumbar standing radiographs. Yonsei Med J 2012;53:1183-1189
  27. Goldberg CJ, Moore DP, Fogarty EE, Dowling FE. Left thoracic curve patterns and their association with disease. Spine (Phila Pa 1976) 1999;24:1228-1233
  28. Wu L, Qiu Y, Wang B, Zhu ZZ, Ma WW. The left thoracic curve pattern: a strong predictor for neural axis abnormalities in patients with "idiopathic" scoliosis. Spine (Phila Pa 1976) 2010;35:182-185