Clustering of electrocardiograph signals in computer-aided Holter analysis

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Long-term electrocardiographic (Holter) signals contain thousands of heartbeats, making manual visual inspection time-consuming and impractical for diagnosis.
• Automated analysis of these complex signals is crucial for efficient and accurate medical decision support.

Purpose of the Study:

• To develop an automated method for reducing the volume of heartbeats in Holter signals without losing diagnostic information.
• To create a computationally efficient process for identifying significant ECG beats for cardiologists.

Main Methods:

• Temporal alignment, feature extraction, and clustering algorithms were applied to electrocardiographic signals.
• Experiments utilized the well-known MIT database, featuring diverse ECG waveforms.
• A comprehensive experimental study was conducted to validate the proposed methods.

Main Results:

• A complete process was developed to extract significant heartbeats from Holter signals.
• The proposed method offers a reasonable computational cost for processing large ECG datasets.
• The identified subset of beats is fully representative of the original signal.

Conclusions:

• The developed method significantly reduces the number of beats requiring cardiologists' attention.
• This tool enhances medical decision support systems by providing a manageable yet informative selection of ECG beats.
• The automated approach improves the efficiency of analyzing long-term electrocardiographic data.