Electrocardiogram pattern recognition and analysis based on artificial neural networks and support vector machines: a...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Pattern Recognition

Background:

• Electrocardiogram (ECG) analysis aids clinicians in monitoring and detecting critical cardiac events.
• ECG analysis is expanding into biometrics for human identification, presenting unique challenges and opportunities.
• Existing clinical ECG analysis methods share similarities and differences with biometric applications.

Purpose of the Study:

• To review ECG processing methods from a pattern recognition viewpoint.
• To highlight features crucial for heartbeat classification.
• To discuss popular classifiers like Artificial Neural Networks and Support Vector Machines.

Main Methods:

• Focus on pattern recognition techniques for ECG signal processing.
• Detailed discussion of Artificial Neural Networks (ANNs) and Support Vector Machines (SVMs).
• Mention of other methods including Hidden Markov Models and Kalman Filtering.

Main Results:

• Identified common features used in heartbeat classification.
• Provided an in-depth review of ANNs and SVMs for ECG analysis.
• Acknowledged the applicability of various pattern recognition methods.

Conclusions:

• ECG analysis is a versatile tool with applications in both clinical diagnostics and biometrics.
• Pattern recognition offers a robust framework for developing advanced ECG analysis systems.
• ANNs and SVMs are prominent and effective classifiers for ECG-based tasks.