Reliable peak detection and feature extraction for wireless electrocardiograms.

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Electrocardiogram (ECG) is crucial for diagnosing heart conditions by analyzing electrical activity.
• Current methods rely on manual inspection in clinical settings, missing conditions revealed during patient movement.
• Cardiac conditions often manifest during daily activities outside controlled environments.

Purpose of the Study:

• To dynamically identify and extract prominent ECG features from signals recorded during everyday activities.
• To develop an automated method for analyzing ECGs outside of clinical settings.
• To improve the detection of cardiac conditions in real-world patient environments.

Main Methods:

• Utilized adaptive thresholding and localization techniques for feature extraction.
• Focused on ECG data from subjects performing common activities like sitting, standing, and stair climbing.
• Developed a method for analyzing ECG signals acquired outside of traditional clinical settings.

Main Results:

• Achieved high accuracy in detecting key ECG peaks: R peak (average %), Q and S peaks (92%).
• Successfully detected T and P peaks with average accuracies of 87% and 84%, respectively.
• Demonstrated the feasibility of accurate ECG feature extraction during non-clinical activities.

Conclusions:

• The proposed method enables accurate identification of essential ECG features during daily patient activities.
• This approach facilitates remote and continuous monitoring for cardiovascular conditions.
• The findings support the development of more accessible and effective cardiac diagnostics.