Korean Journal of Radiology

  • 제22권3호
  • /
  • Pages.334-343
  • /
  • 2021
  • /
  • 1229-6929(pISSN)
  • /
  • 2005-8330(eISSN)
대한영상의학회 (The Korean Society of Radiology)
권호 표지
DOI QR코드

DOI QR Code

Performance of Prediction Models for Diagnosing Severe Aortic Stenosis Based on Aortic Valve Calcium on Cardiac Computed Tomography: Incorporation of Radiomics and Machine Learning

  • Nam gyu Kang (Department of Radiology, Research Institute of Radiological Science, Center for Clinical Imaging Data Science, Severance Hospital, Yonsei University College of Medicine) ;
  • Young Joo Suh (Department of Radiology, Research Institute of Radiological Science, Center for Clinical Imaging Data Science, Severance Hospital, Yonsei University College of Medicine) ;
  • Kyunghwa Han (Department of Radiology, Research Institute of Radiological Science, Center for Clinical Imaging Data Science, Severance Hospital, Yonsei University College of Medicine) ;
  • Young Jin Kim (Department of Radiology, Research Institute of Radiological Science, Center for Clinical Imaging Data Science, Severance Hospital, Yonsei University College of Medicine) ;
  • Byoung Wook Choi (Department of Radiology, Research Institute of Radiological Science, Center for Clinical Imaging Data Science, Severance Hospital, Yonsei University College of Medicine)
  • 투고 : 2020.02.06
  • 심사 : 2020.06.04
  • 발행 : 2021.03.01
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

Objective: We aimed to develop a prediction model for diagnosing severe aortic stenosis (AS) using computed tomography (CT) radiomics features of aortic valve calcium (AVC) and machine learning (ML) algorithms. Materials and Methods: We retrospectively enrolled 408 patients who underwent cardiac CT between March 2010 and August 2017 and had echocardiographic examinations (240 patients with severe AS on echocardiography [the severe AS group] and 168 patients without severe AS [the non-severe AS group]). Data were divided into a training set (312 patients) and a validation set (96 patients). Using non-contrast-enhanced cardiac CT scans, AVC was segmented, and 128 radiomics features for AVC were extracted. After feature selection was performed with three ML algorithms (least absolute shrinkage and selection operator [LASSO], random forests [RFs], and eXtreme Gradient Boosting [XGBoost]), model classifiers for diagnosing severe AS on echocardiography were developed in combination with three different model classifier methods (logistic regression, RF, and XGBoost). The performance (c-index) of each radiomics prediction model was compared with predictions based on AVC volume and score. Results: The radiomics scores derived from LASSO were significantly different between the severe AS and non-severe AS groups in the validation set (median, 1.563 vs. 0.197, respectively, p < 0.001). A radiomics prediction model based on feature selection by LASSO + model classifier by XGBoost showed the highest c-index of 0.921 (95% confidence interval [CI], 0.869-0.973) in the validation set. Compared to prediction models based on AVC volume and score (c-indexes of 0.894 [95% CI, 0.815-0.948] and 0.899 [95% CI, 0.820-0.951], respectively), eight and three of the nine radiomics prediction models showed higher discrimination abilities for severe AS. However, the differences were not statistically significant (p > 0.05 for all). Conclusion: Models based on the radiomics features of AVC and ML algorithms may perform well for diagnosing severe AS, but the added value compared to AVC volume and score should be investigated further.

키워드

과제정보

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (2018R1C1B6007251) and a faculty research grant of Yonsei University College of Medicine (6-2018-0041).

참고문헌

  1. Clavel MA, Burwash IG, Pibarot P. Cardiac imaging for assessing low-gradient severe aortic stenosis. JACC Cardiovasc Imaging 2017;10:185-202
  2. Berthelot-Richer M, Pibarot P, Capoulade R, Dumesnil JG, Dahou A, Thebault C, et al. Discordant grading of aortic stenosis severity: echocardiographic predictors of survival benefit associated with aortic valve replacement. JACC Cardiovasc Imaging 2016;9:797-805
  3. Koos R, Mahnken AH, Sinha AM, Wildberger JE, Hoffmann R, Kühl HP. Aortic valve calcification as a marker for aortic stenosis severity: assessment on 16-MDCT. AJR Am J Roentgenol 2004;183:1813-1818
  4. Pawade T, Clavel MA, Tribouilloy C, Dreyfus J, Mathieu T, Tastet L, et al. Computed tomography aortic valve calcium scoring in patients with aortic stenosis. Circ Cardiovasc Imaging 2018;11:e007146
  5. Clavel MA, Pibarot P, Messika-Zeitoun D, Capoulade R, Malouf J, Aggarval S, et al. Impact of aortic valve calcification, as measured by MDCT, on survival in patients with aortic stenosis: results of an international registry study. J Am Coll Cardiol 2014;64:1202-1213
  6. Baumgartner H, Falk V, Bax JJ, De Bonis M, Hamm C, Holm PJ, et al. 2017 ESC/EACTS Guidelines for the management of valvular heart disease. Eur Heart J 2017;38:2739-2791
  7. Abramowitz Y, Jilaihawi H, Pibarot P, Chakravarty T, Kashif M, Kazuno Y, et al. Severe aortic stenosis with low aortic valve calcification: characteristics and outcome following transcatheter aortic valve implantation. Eur Heart J Cardiovasc Imaging 2017;18:639-647
  8. de Santis A, Tarasoutchi F, Araujo Filho JAB, Vieira MC, Nomura CH, Katz M, et al. Topographic pattern of valve calcification: a new determinant of disease severity in aortic valve stenosis. JACC Cardiovasc Imaging 2018;11:1032-1035
  9. Kolossvary M, Park J, Bang JI, Zhang J, Lee JM, Paeng JC, et al. Identification of invasive and radionuclide imaging markers of coronary plaque vulnerability using radiomic analysis of coronary computed tomography angiography. Eur Heart J-Cardiovasc Imaging 2019;20:1250-1258
  10. Kolossvary M, Karady J, Szilveszter B, Kitslaar P, Hoffmann U, Merkely B, et al. Radiomic features are superior to conventional quantitative computed tomographic metrics to identify coronary plaques with napkin-ring sign. Circ Cardiovasc Imaging 2017;10:e006843
  11. Neisius U, El-Rewaidy H, Nakamori S, Rodriguez J, Manning WJ, Nezafat R. Radiomic analysis of myocardial native T1 imaging discriminates between hypertensive heart disease and hypertrophic cardiomyopathy. JACC Cardiovasc Imaging 2019;12:1946-1954
  12. Baessler B, Luecke C, Lurz J, Klingel K, von Roeder M, de Waha S, et al. Cardiac MRI texture analysis of T1 and T2 maps in patients with infarctlike acute myocarditis. Radiology 2018;289:357-365
  13. Nam K, Suh YJ, Han K, Park SJ, Kim YJ, Choi BW. Value of computed tomography radiomic features for differentiation of periprosthetic mass in patients with suspected prosthetic valve obstruction. Circ Cardiovasc Imaging 2019;12:e009496
  14. Park JE, Park SY, Kim HJ, Kim HS. Reproducibility and generalizability in radiomics modeling: possible strategies in radiologic and statistical perspectives. Korean J Radiol 2019;20:1124-1137
  15. Park YW, Choi YS, Ahn SS, Chang JH, Kim SH, Lee SK. Radiomics MRI phenotyping with machine learning to predict the grade of lower-grade gliomas: a study focused on nonenhancing tumors. Korean J Radiol 2019;20:1381-1389
  16. Park SH, Han K. Methodologic guide for evaluating clinical performance and effect of artificial intelligence technology formMedical diagnosis and prediction. Radiology 2018;286:800-809
  17. Eberhard M, Hinzpeter R, Polacin M, Morsbach F, Maisano F, Nietlispach F, et al. Reproducibility of aortic valve calcification scoring with computed tomography - An interplatform analysis. J Cardiovasc Comput Tomogr 2019;13:92-98
  18. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990;15:827-832
  19. Nishimura RA, Otto CM, Bonow RO, Carabello BA, Erwin JP 3rd, Guyton RA, et al. 2014 AHA/ACC guideline for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Thorac Cardiovasc Surg 2014;148:e1-e132
  20. Breiman L. Random forests. Machine learning 2001;45:5-32
  21. Chen T, Guestrin C. XGBoost: a scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining; 2016 Aug; San Francisco, CA, USA: Association for Computing Machinery; p. 785-794
  22. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44:837-845
  23. Fluss R, Faraggi D, Reiser B. Estimation of the Youden Index and its associated cutoff point. Biom J 2005;47:458-472
  24. Shimizu K, Yamamoto M, Koyama Y, Kodama A, Sato H, Kano S, et al. Usefulness of routine aortic valve calcium score measurement for risk stratification of aortic stenosis and coronary artery disease in patients scheduled cardiac multislice computed tomography. Int J Cardiol Heart Vasc 2015;9:95-99
  25. Cueff C, Serfaty JM, Cimadevilla C, Laissy JP, Himbert D, Tubach F, et al. Measurement of aortic valve calcification using multislice computed tomography: correlation with haemodynamic severity of aortic stenosis and clinical implication for patients with low ejection fraction. Heart 2011;97:721-726
  26. Clavel MA, Messika-Zeitoun D, Pibarot P, Aggarwal SR, Malouf J, Araoz PA, et al. The complex nature of discordant severe calcified aortic valve disease grading: new insights from combined Doppler echocardiographic and computed tomographic study. J Am Coll Cardiol 2013;62:2329-2338
  27. Chandra S, Lang RM, Nicolarsen J, Gayat E, Spencer KT, Mor-Avi V, et al. Bicuspid aortic valve: inter-racial difference in frequency and aortic dimensions. JACC Cardiovasc Imaging 2012;5:981-989
  28. Kong WKF, Regeer MV, Poh KK, Yip JW, van Rosendael PJ, Yeo TC, et al. Inter-ethnic differences in valve morphology, valvular dysfunction, and aortopathy between Asian and European patients with bicuspid aortic valve. Eur Heart J 2018;39:1308-1313
  29. Azzalini L, Ghoshhajra BB, Elmariah S, Passeri JJ, Inglessis I, Palacios IF, et al. The aortic valve calcium nodule score (AVCNS) independently predicts paravalvular regurgitation after transcatheter aortic valve replacement (TAVR). J Cardiovasc Comput Tomogr 2014;8:131-140
  30. Khalique OK, Hahn RT, Gada H, Nazif TM, Vahl TP, George I, et al. Quantity and location of aortic valve complex calcification predicts severity and location of paravalvular regurgitation and frequency of post-dilation after balloon-expandable transcatheter aortic valve replacement. JACC Cardiovasc Interv 2014;7:885-894
  31. Ewe SH, Ng AC, Schuijf JD, van der Kley F, Colli A, Palmen M, et al. Location and severity of aortic valve calcium and implications for aortic regurgitation after transcatheter aortic valve implantation. Am J Cardiol 2011;108:1470-1477
  32. Jilaihawi H, Makkar RR, Kashif M, Okuyama K, Chakravarty T, Shiota T, et al. A revised methodology for aortic-valvar complex calcium quantification for transcatheter aortic valve implantation. Eur Heart J Cardiovasc Imaging 2014;15:1324-1332
  33. Hansson NC, Norgaard BL, Barbanti M, Nielsen NE, Yang TH, Tamburino C, et al. The impact of calcium volume and distribution in aortic root injury related to balloon-expandable transcatheter aortic valve replacement. J Cardiovasc Comput Tomogr 2015;9:382-392
  34. Berenguer R, Pastor-Juan MDR, Canales-Vazquez J, Castro-Garcia M, Villas MV, Mansilla Legorburo F, et al. Radiomics of CT features may be nonreproducible and redundant: influence of CT acquisition parameters. Radiology 2018;288:407-415
  35. Sung P, Lee JM, Joo I, Lee S, Kim TH, Ganeshan B. Evaluation of the impact of iterative reconstruction algorithms on computed tomography texture features of the liver parenchyma using the filtration-histogram method. Korean J Radiol 2019;20:558-568
  36. Lee SM, Lee JG, Lee G, Choe J, Do KH, Kim N, et al. CT image conversion among different reconstruction kernels without a sinogram by using a convolutional neural network. Korean J Radiol 2019;20:295-303
  37. Choe J, Lee SM, Do KH, Lee G, Lee JG, Lee SM, et al. Deep learning-based image conversion of CT reconstruction kernels improves radiomics reproducibility for pulmonary nodules or masses. Radiology 2019;292:365-373
  38. Park S, Lee SM, Do KH, Lee JG, Bae W, Park H, et al. Deep learning algorithm for reducing CT slice thickness: effect on reproducibility of radiomic features in lung cancer. Korean J Radiol 2019;20:1431-1440