Time-Frequency Analysis Based Detection of Dysrhythmia in ECG Using Stockwell Transform

Abstract

Dysrhythmia is the abnormality in rhythm of our cardiac activity. Dysrhythmia is mainly caused by the re-entry of the electric impulse resulting abnormal depolarization of the myocardium cells. Sometimes, such activity causes life threatening ailments. Several myocardial diseases have been studied and detected by help of time frequency representation based techniques effectively. So, a powerful tool called Stockwell transform (ST) has been evolved to provide better time-frequency localization. Wavelet transform based methods arose as an challenging tool for the analysis of ECG signals with both the temporal and the frequency resolution levels. In this study, S-Transform based time-frequency analysis is adopted to detect the Dysrhythmia in ECG signal at high frequencies, which is difficult to study by using Continuous Wavelet transform (CWT). The time frequency analysis is performed over 3 frequency ranges namely low frequency (1–15 Hz) zone, mid-frequency (15–80 Hz) zone and high-frequency (>80 Hz) zone and their respective Integrated time-frequency Power (ITFP) are calculated. The patients with Dysrhythmia has higher ITFP in the high frequency zone than the healthy individuals. The accuracy of the detection of Dysrhythmia is found out to be 88.09% using ST whereas CWT method provides only 58.62% detection accuracy.