Real-Time ECG Delineation with Randomly Selected Wavelet Transform Feature and Random Walk Estimation.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Accurate detection of characteristic points in Electrocardiogram (ECG) signals is crucial for diagnosing heart diseases.
• Existing ECG delineation methods face challenges in accuracy and real-time processing capabilities.

Purpose of the Study:

• To propose a novel feature extraction and machine learning scheme for improved ECG delineation.
• To introduce a new feature, randomly selected wavelet transform (RSWT), for effective ECG morphology representation.

Main Methods:

• Utilizing a randomly selected wavelet transform (RSWT) for feature extraction from ECG signals.
• Training a regression tree on the RSWT feature pool to estimate directional probability towards target points.
• Employing a continual random walk in 1D space to derive the final position of target points.

Main Results:

• The proposed RSWT feature demonstrated effective representation of ECG morphology.
• The regression tree and random walk approach achieved reliable detection of characteristic ECG points.
• Evaluation on the QT database showed superior detection accuracy compared to existing studies.

Conclusions:

• The novel RSWT-based machine learning scheme offers enhanced accuracy for ECG delineation.
• The method provides real-time processing capability, making it suitable for clinical applications.
• This approach holds significant potential for improving the diagnosis of heart diseases through ECG analysis.