Real time ECG characteristic point detection with randomly selected signal pair difference (RSSPD) feature and random...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Automated electrocardiogram (ECG) analysis relies heavily on accurate detection of characteristic points.
• Existing methods for ECG feature extraction and point detection face challenges in robustness and accuracy.

Purpose of the Study:

• To develop a novel statistical scheme for the simultaneous detection of all ECG characteristic points.
• To introduce an effective feature representation for ECG morphology.

Main Methods:

• A novel feature, the randomly selected signal pair difference (RSSPD), was designed to represent ECG morphology.
• A multi-class random forest classifier was trained using the RSSPD feature pool for simultaneous characteristic point detection.
• Post-processing was applied to refine the final output.

Main Results:

• The proposed method achieved robust and accurate detection performance on ECG datasets.
• Evaluation on the QT database demonstrated superior detection accuracy compared to other existing studies.

Conclusions:

• The novel statistical scheme and RSSPD feature offer an effective approach for ECG characteristic point detection.
• The method shows significant potential for improving automated ECG analysis and diagnosis.