[Automatic detection model of hypertrophic cardiomyopathy based on deep convolutional neural network]

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Hypertrophic cardiomyopathy (HCM) diagnosis is crucial for sudden cardiac death risk stratification and genetic screening.
• Conventional multi-lead ECG methods have technical limitations.
• Early and accurate detection of HCM is essential for patient management.

Purpose of the Study:

• To develop an automated method for hypertrophic cardiomyopathy (HCM) detection using single-lead electrocardiogram (ECG) signals.
• To leverage deep learning, specifically convolution neural networks (CNNs), for enhanced ECG analysis.
• To provide a tool for efficient, large-scale preliminary screening of HCM.

Main Methods:

• Utilized single-lead ECG signals as input data.
• Implemented R-wave peak detection, signal segmentation, and resampling per heartbeat.
• Developed and optimized a convolution neural network (CNN) model for feature extraction and HCM classification.
• Trained the model on a dataset of 14,459 heartbeats from public PhysioNet databases.

Main Results:

• The optimized CNN model demonstrated high performance in detecting HCM.
• Achieved an accuracy of 95.98%, sensitivity of 98.03%, and specificity of 95.79%.
• The deep learning approach effectively analyzed single-lead ECG data.

Conclusions:

• Deep learning, specifically CNNs, offers an effective method for automated HCM detection from single-lead ECG.
• This approach overcomes limitations of traditional multi-lead ECG methods.
• The developed method has significant potential for assisting clinicians in rapid, large-scale HCM screening.