Zhang Huiqiong,Jia Weijie,Yu Kai,et al.Deep learning based intelligent auscultation of heart sounds in neonates with congenital heart disease[J].Journal of Clinical Pediatric Surgery,,22():642-648.[doi:10.3760/cma.j.cn101785-202207056-008]
Deep learning based intelligent auscultation of heart sounds in neonates with congenital heart disease
- Keywords:
- Heart Diseases; Heart Sounds; Algorithms; Surgical Procedures; Operative; Child
- Abstract:
- Objective To examine the heart sound signals of four common congenital heart diseases of ventricular septal defect, atrial septal defect, patent ductus arteriosus and patent foramen with associated pulmonary hypertension and propose a deep learning-based intelligent auscultation algorithm for automatic classification of heart sound signals.Methods The algorithm in this study was based upon digital signal processing technology of converting one-dimensional temporal signal classification into a two-dimensional image classification and further applying deep neural networks for automatic classification of heart sound signals.A total of 941 heart sound data samples collected from Children’s Hospital of Zhejiang University School of Medicine were employed for training, validation and testing with a ratio of 8:1:1 for training, validation and testing sets respectively.Additionally, 107 heart sound data samples gathered from a clinical screening environment were collected to validate the effectiveness of intelligent auscultation algorithm in real-world clinical applications.Results In this study, discrete wavelet transformation was utilized to denoise the heart sound signals and there was a significant improvement in model performance.Compared to model without denoising, the denoised model achieved notable enhancements in accuracy, sensitivity, specificity and F1 score on testing set with improvements of 15.8%, 32.6%, 11.1% and 27.3% respectively.Furthermore, the authors compared the performance of several common classification neural network models, including Swin_transform, Vit, Mobilenet, Resenet and Vgg, with their respective F1 scores of 0.905, 0.842, 0.687, 0.814 and 0.864.Finally, using Swin_transform model, tests on the external dataset of 107 cases yielded an accuracy of 0.833, a sensitivity of 0.872 and a specificity of 0.801.Conclusion This study highlights the significant impact of noise and neural network structure on the performance of automatic classification models for CHD heart sound signals.Through the application of discrete wavelet transform for denoising heart sound signals, a substantial improvement in model performance is observed.Among various common classification neural network models, Swin_transform model exhibits the best classification performance.Additionally, the validation of intelligent auscultation algorithm on an external dataset of 107 cases demonstrates its effectiveness in real-world clinical applications, yielding favorable accuracy, sensitivity and specificity results.In summary, this study demonstrates the promising potential of deep learning-based intelligent auscultation algorithms for automatic classification of CHD heart sound signals.
References:
[1] van der Linde D, Konings EEM, Slager MA, et al.Birth prevalence of congenital heart disease worldwide:a systematic review and meta-analysis[J].J Am Coll Cardiol, 2011, 58(21):2241-2247.DOI:10.1016/j.jacc.2011.08.025.
[2] GBD 2017 Congenital Heart Disease Collaborators.Global, regional, and national burden of congenital heart disease, 1990-2017:a systematic analysis for the Global Burden of Disease Study 2017[J].Lancet Child Adolesc Health, 2020, 4(3):185-200.DOI:10.1016/S2352-4642(19)30402-X.
[3] Qiu YX, Jiang W, Zhang JY, et al.Using echocardiography in newborn screening for congenital heart disease may reduce missed diagnoses[J].World J Pediatr, 2022, 18(9):629-631.DOI:10.1007/s12519-022-00560-2.
[4] Pan FX, Li JB, Lou HL, et al.Geographical and socioeconomic factors influence the birth prevalence of congenital heart disease:a population-based cross-sectional study in eastern China[J].Curr Probl Cardiol, 2022, 47(11):101341.DOI:10.1016/j.cpcardiol.2022.101341.
[5] Jacobs JP, O’Brien SM, Pasquali SK, et al.Variation in outcomes for risk-stratified pediatric cardiac surgical operations:an analysis of the STS Congenital Heart Surgery Database[J].Ann Thorac Surg, 2012, 94(2):564-572.DOI:10.1016/j.athoracsur.2012.01.105.
[6] Liu XW, Xu WZ, Yu JG, et al.Screening for congenital heart defects:diversified strategies in current China[J].World J Pediatr Surg, 2019, 2(1):e000051.DOI:10.1136/wjps-2019-000051.
[7] 赵趣鸣, 刘芳, 吴琳, 等.危重先天性心脏病新生儿产科医院出院前漏诊情况分析[J].中华儿科杂志, 2017, 55(4):260-266.DOI:10.3760/cma.j.issn.0578-1310.2017.04.006. Zhao QM, Liu F, Wu L, et al.Assessment of undiagnosed critical congenital heart disease before discharge from maternity hospital[J].Chin J Pediatr, 2017, 55(4):260-266.DOI:10.3760/cma.j.issn.0578-1310.2017.04.006.
[8] Aziz S, Khan MU, Alhaisoni M, et al.Phonocardiogram signal processing for automatic diagnosis of congenital heart disorders through fusion of temporal and cepstral features[J].Sensors (Basel), 2020, 20(13):3790.DOI:10.3390/s20133790.
[9] ?lmez T, Dokur Z.Classification of heart sounds using an artificial neural network[J].Pattern Recognit Lett, 2003, 24(1/3):617-629.DOI:10.1016/S0167-8655(02)00281-7.
[10] Ari S, Hembram K, Saha G.Detection of cardiac abnormality from PCG signal using LMS based least square SVM classifier[J].Expert Syst Appl, 2010, 37(12):8019-8026.DOI:10.1016/j.eswa.2010.05.088.
[11] Ning TK, Ning J, Atanasov N, et al.A fast heart sounds detection and heart murmur classification algorithm[C]//2012 IEEE 11th International Conference on Signal Processing, Beijing, China, 2012.Piscataway, NJ:IEEE, 2012:1629-1632.DOI:10.1109/ICoSP.2012.6491892.
[12] Maknickas V, Maknickas A.Recognition of normal-abnormal phonocardiographic signals using deep convolutional neural networks and mel-frequency spectral coefficients[J].Physiol Meas, 2017, 38(8):1671-1684.DOI:10.1088/1361-6579/aa7841.
[13] Potes C, Parvaneh S, Rahman A, et al.Ensemble of feature-based and deep learning-based classifiers for detection of abnormal heart sounds[C]//2016 Computing in Cardiology Conference (CinC), Vancouver, BC, Canada, 2016.Piscataway, NJ:IEEE, 2016:621-624.
[14] Sahidullah M, Saha G.Design, analysis and experimental evaluation of block based transformation in MFCC computation for speaker recognition[J].Speech Commun, 2012, 54(4):543-565.DOI:10.1016/j.specom.2011.11.004.
[15] Liu Z, Lin YT, Cao Y, et al.Swin transformer:hierarchical vision transformer using shifted windows[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada, 2021.Piscataway, NJ:IEEE, 2021:9992-10002.DOI:10.1109/ICCV48922.2021.00986.
[16] Cortes C, Mohri M, Rostamizadeh A.L2 regularization for learning kernels[EB/OL].(2012-05-09).https://doi.org/10.48550/arXiv.1205.2653.DOI:10.48550/arXiv.1205.2653.
[17] Shensa MJ.The discrete wavelet transform:wedding the a trous and Mallat algorithms[J].IEEE Trans Signal Process, 1992, 40(10):2464-2482.DOI:10.1109/78.157290.
[18] 徐玮泽, 俞凯, 徐佳俊, 等.先天性心脏病心音听诊筛查的人工智能技术应用现状[J].浙江大学学报(医学版), 2020, 49(5):548-555.DOI:10.3785/j.issn.1008-9292.2020.10.01. Xu WZ, Yu K, Xu JJ, et al.Current application status of artificial intelligence technology in cardiac auscultation screening for congenital heart disease[J].J Zhejiang Univ (Med Sci), 2020, 49(5):548-555.DOI:10.3785/j.issn.1008-9292.2020.10.01.
[19] Xu WZ, Yu K, Ye JJ, et al.Automatic pediatric congenital heart disease classification based on heart sound signal[J].Artif Intell Med, 2022, 126:102257.DOI:10.1016/j.artmed.2022.102257.
[20] Alafif T, Boulares M, Barnawi A, et al.Normal and abnormal heart rates recognition using transfer learning[C]//202012th International Conference on Knowledge and Systems Engineering (KSE), Can Tho, Vietnam, 2020.Piscataway, NJ:IEEE, 2020:275-280.DOI:10.1109/KSE50997.2020.9287514.
[21] Yang JJ, Yan K, Wang Z, et al.A novel denoising method for partial discharge signal based on improved variational mode decomposition[J].Energies, 2022, 15(21):8167.DOI:10.3390/en15218167.
[22] Eltrass AS.Novel cascade filter design of improved sparse low-rank matrix estimation and kernel adaptive filtering for ECG denoising and artifacts cancellation[J].Biomed Signal Process Control, 2022, 77:103750.DOI:10.1016/j.bspc.2022.103750.
[23] Schlüter J, Gutenbrunner G.Efficient LEAF:a faster learnable audio frontend of questionable use[C]//202230th European Signal Processing Conference (EUSIPCO), Belgrade, Serbia, 2022.Piscataway, NJ:IEEE, 2022:205-208.DOI:10.23919/EUSIPCO55093.2022.9909910.
Memo
收稿日期:2022-07-29。
基金项目:国家自然科学基金面上项目（82270309）；浙江省“尖兵”“领雁”研发攻关计划（2022C03087）
通讯作者:徐玮泽,Email:weizexu@zju.edu.cn