Novel T-wave Detection Technique with Minimal Processing and RR-Interval Based Enhanced Efficiency

T-wave in electrocardiogram (ECG) is a vital wave component and has potential of diagnosing various cardiac disorders. The present work proposes a novel technique for T-wave peak detection using minimal pre-processing and simple root mean square based decision rule.

The technique uses a two-stage median filter and a Savitzky-Golay smoothing filter for pre-processing. P-QRS-complex is removed from the filtered ECG, and T-wave is left as the most prominent wave segment, which can be detected using a root mean square based adaptive threshold. An RR-interval based T-wave peak correction strategy has been proposed which can handle the challenges of morphological variations in the T-wave, thus increases the detection accuracy.

The proposed technique has been substantiated on a standard QT-database. The detection sensitivity = 97.01%, positive predictivity = 99.61%, detection error rate = 3.36%, and accuracy = 96.66% have been achieved.

A T-wave detection technique requiring minimal pre-processing and with simple decision rule has been designed. The noticeably high positive predictivity rate of the proposed technique shows its efficiency to detect T-wave peak.