Deep Learning-Based Algorithm for Detecting Aortic Stenosis Using Electrocardiography

Joon-Myoung Kwon, Soo Youn Lee, Ki-Hyun Jeon, Yeha Lee, Kyung-Hee Kim, Jinsik Park, Byung-Hee Oh, Myong-Mook Lee, Joon-Myoung Kwon, Soo Youn Lee, Ki-Hyun Jeon, Yeha Lee, Kyung-Hee Kim, Jinsik Park, Byung-Hee Oh, Myong-Mook Lee

Abstract

Background Severe, symptomatic aortic stenosis (AS) is associated with poor prognoses. However, early detection of AS is difficult because of the long asymptomatic period experienced by many patients, during which screening tools are ineffective. The aim of this study was to develop and validate a deep learning-based algorithm, combining a multilayer perceptron and convolutional neural network, for detecting significant AS using ECGs. Methods and Results This retrospective cohort study included adult patients who had undergone both ECG and echocardiography. A deep learning-based algorithm was developed using 39 371 ECGs. Internal validation of the algorithm was performed with 6453 ECGs from one hospital, and external validation was performed with 10 865 ECGs from another hospital. The end point was significant AS (beyond moderate). We used demographic information, features, and 500-Hz, 12-lead ECG raw data as predictive variables. In addition, we identified which region had the most significant effect on the decision-making of the algorithm using a sensitivity map. During internal and external validation, the areas under the receiver operating characteristic curve of the deep learning-based algorithm using 12-lead ECG for detecting significant AS were 0.884 (95% CI, 0.880-0.887) and 0.861 (95% CI, 0.858-0.863), respectively; those using a single-lead ECG signal were 0.845 (95% CI, 0.841-0.848) and 0.821 (95% CI, 0.816-0.825), respectively. The sensitivity map showed the algorithm focused on the T wave of the precordial lead to determine the presence of significant AS. Conclusions The deep learning-based algorithm demonstrated high accuracy for significant AS detection using both 12-lead and single-lead ECGs.

Keywords: aortic valve stenosis; deep learning; electrocardiography.

Figures

Figure 1
Figure 1
Study flowchart.
Figure 2
Figure 2
ECG data and artificial intelligence algorithm. AFIB indicates atrial fibrillation; AFL, atrial flutter; QTc, corrected QT interval; and 2D, 2‐dimensional.
Figure 3
Figure 3
Performance of artificial intelligence algorithms for detecting aortic stenosis. AUC indicates area under the receiver operating characteristic curve; CNN, convolutional neural network; MLP, multilayer perceptron; and ROC, receiver operating characteristic.
Figure 4
Figure 4
Sensitivity map for confirming the region associated with prediction of aortic stenosis (AS). The sensitivity map showed the convolutional neural network's (CNN's) region of algorithm attention for determining the presence of AS. The most important region is in yellow, and the least important region is in black. Because the number of filters for the first convolutional layer was 64, the sensitivity map described the region of importance for determining the presence of AS as grade 64. We visualized grade 0 as black and grade 64 as yellow. The sensitivity map showed the initial area of T wave in V2–V5 as the most important region used by the developed CNN algorithm for the decision.

References

    1. Kodali SK, Velagapudi P, Hahn RT, Abbott D, Leon MB. Valvular heart disease in patients ≥80 years of age. J Am Coll Cardiol. 2018;71:2058–2072.
    1. Carabello BA, Paulus WJ. Aortic stenosis. Lancet. 2009;373:956–966.
    1. Lancellotti P, Magne J, Dulgheru R, Clavel M‐A, Donal E, Vannan MA, Chambers J, Rosenhek R, Habib G, Lloyd G, et al. Outcomes of patients with asymptomatic aortic stenosis followed up in heart valve clinics. JAMA Cardiol. 2018;3:1060.
    1. Bonow RO. Exercise hemodynamics and risk assessment in asymptomatic aortic stenosis. Circulation. 2012;126:803–805.
    1. Leon MB, Smith CR, Mack M, Miller DC, Moses JW, Svensson LG, Tuzcu EM, Webb JG, Fontana GP, Makkar RR, et al. Transcatheter aortic‐valve implantation for aortic stenosis in patients who cannot undergo surgery. N Engl J Med. 2010;363:1597–1607.
    1. Ben‐Dor I, Pichard AD, Gonzalez MA, Weissman G, Li Y, Goldstein SA, Okubagzi P, Syed AI, Maluenda G, Collins SD, et al. Correlates and causes of death in patients with severe symptomatic aortic stenosis who are not eligible to participate in a clinical trial of transcatheter aortic valve implantation. Circulation. 2010;122:S37–S42.
    1. Maes F, Lerakis S, Barbosa Ribeiro H, Gilard M, Cavalcante JL, Makkar R, Herrmann HC, Windecker S, Enriquez‐Sarano M, Cheema AN, et al. Outcomes from transcatheter aortic valve replacement in patients with low‐flow, low‐gradient aortic stenosis and left ventricular ejection fraction less than 30%. JAMA Cardiol. 2019;4:64.
    1. Baumgartner H, Falk V, Bax JJ, De Bonis M, Hamm C, Holm PJ, Iung B, Lancellotti P, Lansac E, Rodriguez Muñoz D, et al. 2017 ESC/EACTS guidelines for the management of valvular heart disease. Eur Heart J. 2017;38:2739–2791.
    1. Saikrishnan N, Kumar G, Sawaya FJ, Lerakis S, Yoganathan AP. Accurate assessment of aortic stenosis. Circulation. 2014;129:244–253.
    1. Maganti K, Rigolin VH, Sarano ME, Bonow RO. Valvular heart disease: diagnosis and management. Mayo Clin Proc. 2010;85:483–500.
    1. Alley W, Mahler S. Chapter 54: valvular emergency In: Tintinalli JE, Stapczynski JS, Ma OJ, Yealy DM, Meckler GD, Cline D, eds. Tintinalli's Emergency Medicine: A Comprehensive Study Guide. New York, NY: McGraw Hill Education; 2016;373–380.
    1. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521:436–444.
    1. Gulshan V, Peng L, Coram M, Stumpe MC, Wu D, Narayanaswamy A, Venugopalan S, Widner K, Madams T, Cuadros J, et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA. 2016;304:649–656.
    1. Attia ZI, Kapa S, Lopez‐Jimenez F, McKie PM, Ladewig DJ, Satam G, Pellikka PA, Enriquez‐Sarano M, Noseworthy PA, Munger TM, et al. Screening for cardiac contractile dysfunction using an artificial intelligence–enabled electrocardiogram. Nat Med. 2019;25:70–74.
    1. Kwon J‐M, Lee Y, Lee Y, Lee S, Park J. An algorithm based on deep learning for predicting in‐hospital cardiac arrest. J Am Heart Assoc. 2018;7:e008678 DOI: 10.1161/JAHA.118.008678.
    1. Attia ZI, Noseworthy PA, Lopez‐Jimenez F, Asirvatham SJ, Deshmukh AJ, Gersh BJ, Carter RE, Yao X, Rabinstein AA, Erickson BJ, et al. An artificial intelligence‐enabled ECG algorithm for the identification of patients with atrial fibrillation during sinus rhythm: a retrospective analysis of outcome prediction. Lancet. 2019;394:861–867.
    1. Bonow RO, Brown AS, Gillam LD, Kapadia SR, Kavinsky CJ, Lindman BR, Mack MJ, Thourani VH. ACC/AATS/AHA/ASE/EACTS/HVS/SCA/SCAI/SCCT/SCMR/STS 2017 appropriate use criteria for the treatment of patients with severe aortic stenosis. J Am Coll Cardiol. 2017;70:2566–2598.
    1. Bengio Y, Courville A, Vincent P. Representation learning: a review and new perspectives. IEEE Trans Pattern Anal Mach Intell. 2013;35:1798–1828.
    1. Schmidhuber J. Deep learning in neural networks: an overview. Neural Networks. 2015;61:85–117.
    1. Hannun AY, Rajpurkar P, Haghpanahi M, Tison GH, Bourn C, Turakhia MP, Ng AY. Cardiologist‐level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network. Nat Med. 2019;25:65–69.
    1. He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. Proc IEEE Int Conf Comput Vis. 2016;1:770–778.
    1. LeCun Y, Boser B, Denker JS, Howard RE, Habbard W, Jackel LD. Handwritten digit recognition with a back‐propagation network. Adv Neural Inf Process Syst. 1990;1:396–404.
    1. LeCun Y, Bottou L, Bengio Y, Haffner P. Gradient‐based learning applied to document recognition. Proc IEEE. 1998;1:2278–2324.
    1. Dietterich TG. Ensemble methods in machine learning. In: Multiple Classifier System. London, UK: Springer; 2000;1–15.
    1. Abadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, Devin M, Ghemawat S, Irving G, Isard M, et al. TensorFlow: a system for large‐scale machine learning TensorFlow: a system for large‐scale machine learning. 12th USENIX Symp Oper Syst Des Implement (OSDI ‘16). 2016;265–284.
    1. Pandit SV. ECG baseline drift removal through STFT. In: Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE; 1:1405–1406.
    1. Kuhn M, Johnson K. Applied Predictive Modeling. New York, NY: Springer; 2013.
    1. Mortazavi BJ, Downing NS, Bucholz EM, Dharmarajan K, Manhapra A, Li SX, Negahban SN, Krumholz HM. Analysis of machine learning techniques for heart failure readmissions. Circ Cardiovasc Qual Outcomes. 2016;9:629–640.
    1. Frizzell JD, Liang L, Schulte PJ, Yancy CW, Heidenreich PA, Hernandez AF, Bhatt DL, Fonarow GC, Laskey WK. Prediction of 30‐day all‐cause readmissions in patients hospitalized for heart failure. JAMA Cardiol. 2017;2:204.
    1. Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad‐CAM: visual explanations from deep networks via gradient‐based localization. Proc IEEE Int Conf Comput Vis. 2017;1:618–626.
    1. Fawcett T. An introduction to ROC analysis. Pattern Recognit Lett. 2006;27:861–874.
    1. Carpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med. 2000;19:1141–1164.
    1. Russo R, Rizzoli G, Stritoni P, Seminara G, Rubino M, Brumana T. T‐wave changes in patients with hemodynamic evidence of systolic or diastolic overload of the left ventricle: a retrospective study on 168 patients with isolated chronic aortic valve disease. Int J Cardiol. 1987;14:137–143.
    1. Xin D, Zheng W, Xuyu J, Yuemin S, Wenjuan Z, Bo B, Xuefang Y, Canliang H. Changes of the electrocardiographic strain pattern in patients with aortic stenosis and its underling mechanisms. Heart. 2011;97:A239–A240.
    1. Greve AM, Gerdts E, Boman K, Gohlke‐Baerwolf C, Rossebø AB, Hammer‐Hansen S, Køber L, Willenheimer R, Wachtell K. Differences in cardiovascular risk profile between electrocardiographic hypertrophy versus strain in asymptomatic patients with aortic stenosis (from SEAS Data). Am J Cardiol. 2011;108:541–547.
    1. Greve AM, Boman K, Gohlke‐Baerwolf C, Kesäniemi YA, Nienaber C, Ray S, Egstrup K, Rossebø AB, Devereux RB, Køber L, et al. Clinical implications of electrocardiographic left ventricular strain and hypertrophy in asymptomatic patients with aortic stenosis. Circulation. 2012;125:346–353.
    1. Hering D, Piper C, Horstkotte D. Influence of atypical symptoms and electrocardiographic signs of left ventricular hypertrophy or ST‐segment/T‐wave abnormalities on the natural history of otherwise asymptomatic adults with moderate to severe aortic stenosis: preliminary communication. J Heart Valve Dis. 2004;13:182–187.
    1. Vranic II. Electrocardiographic appearance of aortic stenosis before and after aortic valve replacement. Ann Noninvasive Electrocardiol. 2017;22:e12457.
    1. Xiao HB, Ramzy IS, Bowker TJ, Dancy M. Electrocardiographic associations of right precordial Q waves help to distinguish anterior myocardial infarction from aortic stenosis. Int J Cardiol. 2002;82:159–166.
    1. Taniguchi T, Shiomi H, Kosuge M, Morimoto T, Nakatsuma K, Nishiga M, Sasa T, Saito N, Kimura T. Prognostic significance of ST‐segment elevation in leads V1–2 in patients with severe aortic stenosis. Circ J. 2016;80:526–534.
    1. Pisano ED, Gatsonis C, Hendrick E, Yaffe M, Baum JK, Acharyya S, Conant EF, Fajardo LL, Bassett L, D'Orsi C, et al. Diagnostic performance of digital versus film mammography for breast‐cancer screening. N Engl J Med. 2005;353:1773–1783.
    1. Haug U, Kuntz KM, Knudsen AB, Hundt S, Brenner H. Sensitivity of immunochemical faecal occult blood testing for detecting left‐ vs right‐sided colorectal neoplasia. Br J Cancer. 2011;104:1779–1785.
    1. Schlegel TT, Kulecz WB, Feiveson AH, Greco EC, DePalma JL, Starc V, Vrtovec B, Rahman MA, Bungo MW, Hayat MJ, et al. Accuracy of advanced versus strictly conventional 12‐lead ECG for detection and screening of coronary artery disease, left ventricular hypertrophy and left ventricular systolic dysfunction. BMC Cardiovasc Disord. 2010;10:28.
    1. Chen X, Duan Y, Houthooft R, Schulman J, Sutskever I, Abbeel P. InfoGAN: interpretable representation learning by information maximizing generative adversarial nets. Neural Inf Process Syst. 2016;1:2172–2180.
    1. Fong RC, Vedaldi A. Interpretable explanations of black boxes by meaningful perturbation. Proc IEEE Int Conf Comput Vis. 2017;1:3449–3457.

Source: PubMed

Sponzoři a spolupracovníci

Zdravotní podmínky

Drogové intervence

3
Předplatit