Correlation Method for Detection of Transient T-Wave Alternans in Digital ECG Recordings

Background: Spectral and complex demodulation methods for detection of microvolt T-wave alternans (TWA) have limited ability to identify transient (nonstationary) TWA episodes. We aimed to develop and test new time-domain technique allowing TWA detection and quantification in as few as 7 consecutive beats from sinus rhythm ECGs. Methods and Results: Quantification of TWA during sinus rhythm required preprocessing consisting of: low-pass filtering, RR stability testing, baseline and respiratory modulation removal, and T-wave windowing and synchronization. Our time-domain correlation method (CM) detects TWA by computing, for each consecutive T wave, an alternans correlation index based on a cross-correlation technique. CM allows quantitative analysis of the amplitude, duration, and overall magnitude of the TWA episode. The technical performance of CM was confirmed in testing with simulated TWA of varying amplitude, duration, and noisy conditions. The clinical performance of CM was demonstrated by analyzing digital Holter recordings of 39 long QT syndrome patients compared to 36 healthy subjects. CM identified TWA in 17 (44%) patients with nonstationary TWA detected in 8. Conclusion: Our computer algorithms consisting of ECG preprocessing and TWA quantification by the correlation method provides the opportunity to detect nonstationary and stationary TWA in sinus rhythm of digital Holter ECG recordings.