Korean Journal of Radiology

The Korean Society of Radiology (대한영상의학회)
권호 표지
DOI QR코드

DOI QR Code

Automated Segmentation of Left Ventricular Myocardium on Cardiac Computed Tomography Using Deep Learning

  • Hyun Jung Koo (Department of Radiology and Research Institute of Radiology, Asan Medical Center, University of Ulsan College of Medicine) ;
  • June-Goo Lee (Biomedical Engineering Research Center, Asan Institute of Life Sciences, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Ji Yeon Ko (Biomedical Engineering Research Center, Asan Institute of Life Sciences, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Gaeun Lee (Biomedical Engineering Research Center, Asan Institute of Life Sciences, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Joon-Won Kang (Department of Radiology and Research Institute of Radiology, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Young-Hak Kim (Division of Cardiology, Internal Medicine, Asan Medical Center, University of Ulsan College of Medicine) ;
  • Dong Hyun Yang (Department of Radiology and Research Institute of Radiology, Asan Medical Center, University of Ulsan College of Medicine)
  • Received : 2019.08.27
  • Accepted : 2020.02.09
  • Published : 2020.06.01
Download PDF
⟨ Previous Next ⟩

Abstract

Objective: To evaluate the accuracy of a deep learning-based automated segmentation of the left ventricle (LV) myocardium using cardiac CT. Materials and Methods: To develop a fully automated algorithm, 100 subjects with coronary artery disease were randomly selected as a development set (50 training / 20 validation / 30 internal test). An experienced cardiac radiologist generated the manual segmentation of the development set. The trained model was evaluated using 1000 validation set generated by an experienced technician. Visual assessment was performed to compare the manual and automatic segmentations. In a quantitative analysis, sensitivity and specificity were calculated according to the number of pixels where two three-dimensional masks of the manual and deep learning segmentations overlapped. Similarity indices, such as the Dice similarity coefficient (DSC), were used to evaluate the margin of each segmented masks. Results: The sensitivity and specificity of automated segmentation for each segment (1-16 segments) were high (85.5-100.0%). The DSC was 88.3 ± 6.2%. Among randomly selected 100 cases, all manual segmentation and deep learning masks for visual analysis were classified as very accurate to mostly accurate and there were no inaccurate cases (manual vs. deep learning: very accurate, 31 vs. 53; accurate, 64 vs. 39; mostly accurate, 15 vs. 8). The number of very accurate cases for deep learning masks was greater than that for manually segmented masks. Conclusion: We present deep learning-based automatic segmentation of the LV myocardium and the results are comparable to manual segmentation data with high sensitivity, specificity, and high similarity scores.

Keywords

Acknowledgement

This work was supported by National Research Foundation of Korea (NRF) grants funded by the Korean government (MSIP) (NRF-2020R1A2C2003843) and a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (HI18C0022).

References

  1. Lin FY, Devereux RB, Roman MJ, Meng J, Jow VM, Jacobs A, et al. Cardiac chamber volumes, function, and mass as determined by 64-multidetector row computed tomography: mean values among healthy adults free of hypertension and obesity. JACC Cardiovasc Imaging 2008;1:782-786  https://doi.org/10.1016/j.jcmg.2008.04.015
  2. Juergens KU, Seifarth H, Range F, Wienbeck S, Wenker M, Heindel W, et al. Automated threshold-based 3D segmentation versus short-axis planimetry for assessment of global left ventricular function with dual-source MDCT. AJR Am J Roentgenol 2008;190:308-314  https://doi.org/10.2214/AJR.07.2283
  3. Oktay O, Ferrante E, Kamnitsas K, Heinrich M, Bai W, Caballero J, et al. Anatomically Constrained Neural Networks (ACNNs): application to cardiac image enhancement and segmentation. IEEE Trans Med Imaging 2018;37:384-395  https://doi.org/10.1109/TMI.2017.2743464
  4. Leclerc S, Smistad E, Pedrosa J, Ostvik A, Cervenansky F, Espinosa F, et al. Deep learning for segmentation using an open large-scale dataset in 2D echocardiography. IEEE Trans Med Imaging 2019;38:2198-2210  https://doi.org/10.1109/TMI.2019.2900516
  5. Mahnken AH, Muhlenbruch G, Koos R, Stanzel S, Busch PS, Niethammer M, et al. Automated vs. manual assessment of left ventricular function in cardiac multidetector row computed tomography: comparison with magnetic resonance imaging. Eur Radiol 2006;16:1416-1423  https://doi.org/10.1007/s00330-006-0226-1
  6. Emad O, Yassine IA, Fahmy AS. Automatic localization of the left ventricle in cardiac MRI images using deep learning. Conf Proc IEEE Eng Med Biol Soc 2015;2015:683-686 
  7. Plumhans C, Keil S, Ocklenburg C, Muhlenbruch G, Behrendt FF, Gunther RW, et al. Comparison of manual, semi- and fully automated heart segmentation for assessing global left ventricular function in multidetector computed tomography. Invest Radiol 2009;44:476-482  https://doi.org/10.1097/RLI.0b013e3181aaf4e1
  8. Lorenz C, von Berg J. A comprehensive shape model of the heart. Med Image Anal 2006;10:657-670  https://doi.org/10.1016/j.media.2006.03.004
  9. Lehnert T, Wrzesniak A, Bernhardt D, Ackermann H, Kerl JM, Vega-Higuera F, et al. Fully automated right ventricular volumetry from ECG-gated coronary CT angiography data: evaluation of prototype software. Int J Cardiovasc Imaging 2013;29:489-496  https://doi.org/10.1007/s10554-012-0109-2
  10. Ecabert O, Peters J, Walker MJ, Ivanc T, Lorenz C, von Berg J, et al. Segmentation of the heart and great vessels in CT images using a model-based adaptation framework. Med Image Anal 2011;15:863-876  https://doi.org/10.1016/j.media.2011.06.004
  11. Coche E, Walker MJ, Zech F, de Crombrugghe R, Vlassenbroek A. Quantitative right and left ventricular functional analysis during gated whole-chest MDCT: a feasibility study comparing automatic segmentation to semi-manual contouring. Eur J Radiol 2010;74:e138-e143  https://doi.org/10.1016/j.ejrad.2009.05.037
  12. Bauer RW, Kraus B, Bernhardt D, Kerl JM, Lehnert T, Ackermann H, et al. Computer-based automated left atrium segmentation and volumetry from ECG-gated coronary CT angiography data: comparison with manual slice segmentation and ultrasound planimetric methods. Rofo 2010;182:1110-1117  https://doi.org/10.1055/s-0029-1245729
  13. Abadi S, Roguin A, Engel A, Lessick J. Feasibility of automatic assessment of four-chamber cardiac function with MDCT: initial clinical application and validation. Eur J Radiol 2010;74:175-181  https://doi.org/10.1016/j.ejrad.2009.01.035
  14. Yang DH, Kang SJ, Koo HJ, Kweon J, Kang JW, Lim TH, et al. Incremental value of subtended myocardial mass for identifying FFR-verified ischemia using quantitative CT angiography: comparison with quantitative coronary angiography and CT-FFR. JACC Cardiovasc Imaging 2019;12:707-717  https://doi.org/10.1016/j.jcmg.2017.10.027
  15. Kang SJ, Kim YH, Lee JG, Kang DY, Lee PH, Ahn JM, et al. Impact of subtended myocardial mass assessed by coronary computed tomographic angiography-based myocardial segmentation. Am J Cardiol 2019;123:757-763  https://doi.org/10.1016/j.amjcard.2018.11.042
  16. Weustink AC, Mollet NR, Pugliese F, Meijboom WB, Nieman K, Heijenbrok-Kal MH, et al. Optimal electrocardiographic pulsing windows and heart rate: effect on image quality and radiation exposure at dual-source coronary CT angiography. Radiology 2008;248:792-798  https://doi.org/10.1148/radiol.2483072098
  17. Shelhamer E, Long J, Darrell T. Fully convolutional networks for semantic segmentation. IEEE Trans Pattern Anal Mach Intell 2017;39:640-651  https://doi.org/10.1109/TPAMI.2016.2572683
  18. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition [updated Apr 2015]. Cornell University, 2014. Available at: https://arxiv.org/abs/1409.1556 
  19. Hamer AW, Takayama M, Abraham KA, Roche AH, Kerr AR, Williams BF, et al. End-systolic volume and long-term survival after coronary artery bypass graft surgery in patients with impaired left ventricular function. Circulation 1994;90:2899-2904  https://doi.org/10.1161/01.CIR.90.6.2899
  20. White HD, Norris RM, Brown MA, Brandt PW, Whitlock RM, Wild CJ. Left ventricular end-systolic volume as the major determinant of survival after recovery from myocardial infarction. Circulation 1987;76:44-51  https://doi.org/10.1161/01.CIR.76.1.44
  21. Fischbach R, Juergens KU, Ozgun M, Maintz D, Grude M, Seifarth H, et al. Assessment of regional left ventricular function with multidetector-row computed tomography versus magnetic resonance imaging. Eur Radiol 2007;17:1009-1017  https://doi.org/10.1007/s00330-006-0438-4
  22. Sugeng L, Mor-Avi V, Weinert L, Niel J, Ebner C, Steringer-Mascherbauer R, et al. Quantitative assessment of left ventricular size and function: side-by-side comparison of real-time three-dimensional echocardiography and computed tomography with magnetic resonance reference. Circulation 2006;114:654-661  https://doi.org/10.1161/CIRCULATIONAHA.106.626143
  23. Heuschmid M, Rothfuss JK, Schroeder S, Fenchel M, Stauder N, Burgstahler C, et al. Assessment of left ventricular myocardial function using 16-slice multidetector-row computed tomography: comparison with magnetic resonance imaging and echocardiography. Eur Radiol 2006;16:551-559  https://doi.org/10.1007/s00330-005-0015-2
  24. Lessick J, Mutlak D, Rispler S, Ghersin E, Dragu R, Litmanovich D, et al. Comparison of multidetector computed tomography versus echocardiography for assessing regional left ventricular function. Am J Cardiol 2005;96:1011-1015  https://doi.org/10.1016/j.amjcard.2005.05.062
  25. Grude M, Juergens KU, Wichter T, Paul M, Fallenberg EM, Muller JG, et al. Evaluation of global left ventricular myocardial function with electrocardiogram-gated multidetector computed tomography: comparison with magnetic resonance imaging. Invest Radiol 2003;38:653-661  https://doi.org/10.1097/01.rli.0000077070.40713.76
  26. Dirksen MS, Bax JJ, de Roos A, Jukema JW, van der Geest RJ, Geleijns K, et al. Usefulness of dynamic multislice computed tomography of left ventricular function in unstable angina pectoris and comparison with echocardiography. Am J Cardiol 2002;90:1157-1160  https://doi.org/10.1016/S0002-9149(02)02789-3
  27. Belge B, Coche E, Pasquet A, Vanoverschelde JL, Gerber BL. Accurate estimation of global and regional cardiac function by retrospectively gated multidetector row computed tomography: comparison with cine magnetic resonance imaging. Eur Radiol 2006;16:1424-1433  https://doi.org/10.1007/s00330-006-0169-6
  28. Yamamuro M, Tadamura E, Kubo S, Toyoda H, Nishina T, Ohba M, et al. Cardiac functional analysis with multi-detector row CT and segmental reconstruction algorithm: comparison with echocardiography, SPECT, and MR imaging. Radiology 2005;234:381-390  https://doi.org/10.1148/radiol.2342031271
  29. Schlosser T, Pagonidis K, Herborn CU, Hunold P, Waltering KU, Lauenstein TC, et al. Assessment of left ventricular parameters using 16-MDCT and new software for endocardial and epicardial border delineation. AJR Am J Roentgenol 2005;184:765-773  https://doi.org/10.2214/ajr.184.3.01840765
  30. Krishnam MS, Tomasian A, Iv M, Ruehm SG, Saleh R, Panknin C, et al. Left ventricular ejection fraction using 64-slice CT coronary angiography and new evaluation software: initial experience. Br J Radiol 2008;81:450-455 https://doi.org/10.1259/bjr/54748900