Multimodal ECG heartbeat classification method based on a convolutional neural network embedded with FCA.

Area of Science:

• Biomedical Engineering
• Medical Informatics
• Cardiology

Background:

• Arrhythmias are irregular heart rhythms requiring accurate diagnosis.
• Automated electrocardiogram (ECG) signal classification is crucial for predicting arrhythmias.
• Existing methods primarily analyze 1D ECG signals, limiting comprehensive analysis.

Purpose of the Study:

• To develop an advanced method for classifying arrhythmias by fusing multiple ECG signal modalities.
• To enhance the accuracy and reliability of automated arrhythmia detection systems.

Main Methods:

• ECG signals were transformed into modal images using Recurrence Plot (RP), Gram Field (GAF), and Markov Transition Field (MTF).
• A Convolutional Neural Network (CNN) model integrated with Feature Channel Attention (FCA) was employed for multimodal ECG image classification.
• The proposed model was evaluated on the MIT-BIH arrhythmia database, classifying five types of arrhythmias.

Main Results:

• The multimodal ECG classification model achieved an accuracy of 99.6% on the MIT-BIH arrhythmia database.
• The CNN-based model with FCA demonstrated superior performance compared to previous state-of-the-art models.
• The experimental results validated the reliability and effectiveness of the proposed approach for arrhythmia classification.

Conclusions:

• Fusing multiple ECG signal modalities into images significantly enhances arrhythmia classification accuracy.
• The developed CNN-based model with FCA offers a robust and reliable solution for automated arrhythmia detection.
• This multimodal approach represents a promising advancement in diagnosing and predicting cardiac arrhythmias.