Myocardial Infarction Detection Based on Multi-lead Ensemble Neural Network.

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Myocardial infarction (MI) detection via electrocardiogram (ECG) is critical for patient survival.
• Accurate differentiation of MI subtypes, such as anterior (AMI) and inferior (IMI), remains a challenge.

Purpose of the Study:

• To develop and evaluate a multi-lead ensemble neural network (MENN) for automated detection of AMI and IMI.
• To leverage multi-lead ECG signals and ensemble learning for enhanced classification performance.

Main Methods:

• A novel MENN architecture combining three sub-networks was proposed.
• The model utilized multi-lead ECG signals to capture comprehensive cardiac electrical activity.
• Performance was assessed using 5-fold inter-subject cross-validation on the PTB database.

Main Results:

• For AMI detection, the MENN achieved 98.35% sensitivity, 97.49% specificity, and 97.92% AUC.
• For IMI detection, the MENN achieved 93.17% sensitivity, 92.02% specificity, and 92.60% AUC.
• The proposed method demonstrated state-of-the-art results, outperforming existing baseline methods.

Conclusions:

• The MENN shows significant potential for accurate and automated MI diagnosis.
• The integration of multi-lead ECG data and ensemble learning enhances diagnostic capabilities.
• This approach could lead to improved clinical outcomes for MI patients.