Detection of QRS complexes in electrocardiogram using support vector machine

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Electrocardiogram (ECG) analysis is crucial for diagnosing cardiac conditions.
• Accurate detection of QRS complexes is fundamental for automated ECG interpretation.
• Existing QRS detection methods face challenges with noise and signal variability.

Purpose of the Study:

• To develop and evaluate a support vector machine (SVM) based algorithm for QRS complex detection in ECG signals.
• To compare the performance of single-lead versus 12-lead ECG analysis for QRS detection.
• To assess the impact of data preprocessing and classifier parameters on detection accuracy.

Main Methods:

• Digital filtering techniques were employed to preprocess ECG signals, removing noise and baseline wander.
• A support vector machine (SVM) was utilized as a classifier to distinguish QRS complexes from other ECG segments.
• Two distinct algorithms were developed: one for single-lead ECG and another for simultaneous 12-lead ECG analysis.
• The algorithms were validated using the standard CSE ECG database.

Main Results:

• The single-lead ECG algorithm achieved a QRS detection rate of 99.3%.
• The 12-lead ECG algorithm demonstrated a superior detection rate of 99.75%.
• False negative rates decreased from 0.7% (single-lead) to 0.26% (12-lead).
• False positive rates significantly reduced from 12.4% (single-lead) to 1.61% (12-lead).

Conclusions:

• Support vector machines provide an effective method for QRS complex detection in ECG.
• Simultaneous analysis of 12-lead ECG data yields significantly higher accuracy and fewer errors than single-lead analysis.
• Optimal performance is contingent upon careful selection of training data, data representation, and classifier parameters.