Detection of ST segment deviation episodes in ECG using KLT with an ensemble neural classifier

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Coronary heart disease (CHD) diagnosis relies on accurate electrocardiogram (ECG) analysis.
• Ambulatory ECG monitoring allows for continuous patient assessment but requires robust automated analysis techniques.
• Detecting ST segment deviations is crucial for identifying myocardial ischemia.

Purpose of the Study:

• To develop and validate an automated technique for detecting ST segment deviations in ambulatory ECG recordings.
• To improve the accuracy and efficiency of ischemia detection for coronary heart disease diagnosis.
• To assess the potential of the proposed method for practical clinical application.

Main Methods:

• Preprocessing of ambulatory ECG signals including noise filtering, baseline removal, and QRS complex detection using discrete wavelet transform (DWT).
• Dimensionality reduction of ST segment data utilizing lead-dependent Karhunen-Loève transform (KLT) bases.
• Classification of ST deviation episodes using an ensemble of backpropagation neural networks.

Main Results:

• The proposed method demonstrated high performance in detecting ST segment deviations.
• Achieved sensitivity of 90.75% and positive predictive value of 89.2%.
• Results compare favorably with existing research in the field.

Conclusions:

• The developed technique offers a reliable approach for automated ST segment deviation detection.
• The method shows significant potential for integration into practical ischemia detection systems.
• This automated approach can aid in the timely diagnosis of coronary heart disease.