Deep Learning for Heart Sound Abnormality of Infants: Proof-of-Concept Study of 1D and 2D Representations
Open Access
Abstract
Introduction: Advanced identification and intervention for Congenital Heart Defects (CHDs) in pediatric populations are crucial, as approximately 1% of neonates worldwide present with these conditions. Traditional methods of diagnosing CHDs often rely on stethoscope auscultation, which heavily depends on the clinician’s expertise and may lead to the oversight of subtle acoustic indicators. Objectives: This study introduces an innovative deep-learning framework designed for the early diagnosis of congenital heart disease. It utilizes time-series data obtained from cardiac auditory signals captured through stethoscopes. Methods: The audio signals were processed into time–frequency representations using Mel-Frequency Cepstral Coefficients (MFCCs). The architecture of the model combines Convolutional Neural Networks (CNNs) for effective feature extraction with Long Short-Term Memory (LSTM) networks to accurately model temporal dependencies. Impressively, the model achieved an accuracy of 98.91% in the early detection of CHDs. Results: While traditional diagnostic tools like Electrocardiograms (ECG) and Phonocardiograms (PCG) remain indispensable for confirming diagnoses, many AI studies have primarily targeted ECG and PCG datasets. This approach emphasizes the potential of cardiac acoustics for the early diagnosis of CHDs, which could lead to improved clinical outcomes for infants. Notably, the dataset used in this research is publicly available, enabling wider application and model training within the research community.Keywords
All Related Versions
Funding Information
- Institute of Information & Communications Technology Planning & Evaluation (IITP) (IITP-2023-RS-2023-00256629, 23-CM-DI-11)
- Institute of Civil-Military Technology Cooperation (IITP-2023-RS-2023-00256629, 23-CM-DI-11)
This publication has 26 references indexed in Scilit:
- PCTMF-Net: heart sound classification with parallel CNNs-transformer and second-order spectral analysisThe Visual Computer, 2023
- Heart Sound Classification Using Deep Learning Techniques Based on Log-mel SpectrogramCircuits, Systems, and Signal Processing, 2022
- Deep learning-based computer-aided heart sound analysis in children with left-to-right shunt congenital heart diseaseInternational Journal of Cardiology, 2022
- Hidden Markov model-based heartbeat detector using electrocardiogram and arterial pressure signalsBiomedical Engineering Letters, 2021
- Toward improving ECG biometric identification using cascaded convolutional neural networksNeurocomputing, 2020
- An efficient heart sound segmentation approach using kurtosis and zero frequency filter featuresBiomedical Signal Processing and Control, 2020
- Phonocardiogram signals processing approach for PASCAL Classifying Heart Sounds ChallengeSignal, Image and Video Processing, 2018
- An open access database for the evaluation of heart sound algorithmsPhysiological Measurement, 2016
- Classification of Heart Sound Signals Using Multi-modal FeaturesProcedia Computer Science, 2015
- A novel method for pediatric heart sound segmentation without using the ECGComputer Methods and Programs in Biomedicine, 2010