Myocardial infarction detection method based on the continuous T-wave area feature and multi-lead-fusion deep...

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Myocardial infarction (MI) poses a significant threat to cardiovascular health.
• Early and accurate detection of MI is crucial for effective treatment and patient outcomes.
• Electrocardiogram (ECG) is a primary diagnostic tool for cardiac conditions.

Purpose of the Study:

• To develop and validate an autonomous algorithm for classifying myocardial infarction (MI) using electrocardiogram (ECG) data.
• To explore a novel feature extraction method combining T-wave characteristics and deep learning.
• To improve the accuracy and reliability of MI detection through feature fusion.

Main Methods:

• A novel method fusing continuous T-wave area (C_TWA) features with ECG deep features was developed.
• C_TWA features were extracted using an adaptive sliding window method to analyze T-wave variability.
• A multi-lead fusion convolutional neural network was employed for deep feature extraction, followed by soft attention fusion and multi-layer perceptron classification.

Main Results:

• The proposed method achieved high performance on the PTB dataset, with 97.67% accuracy, 96.59% precision, and 98.96% recall.
• On a clinical dataset, the method demonstrated strong results with 93.15% accuracy, 93.20% precision, and 95.14% recall.
• The fusion of C_TWA features and deep ECG features significantly improved detection accuracy.

Conclusions:

• The developed algorithm effectively extracts C_TWA features and integrates them with deep ECG signal features.
• This approach enhances the accuracy of myocardial infarction detection, showing promising results on clinical datasets.
• The study highlights the potential of AI-driven ECG analysis for improved cardiovascular diagnostics.