T-wave alternans identification in routine exercise ECG tracings: Comparison of methods

T-wave alternans (TWA) is often measured in special exercise ECGs specifically made for TWA testing in order to minimize the noise level. Still, TWA utility in clinical evaluation would be magnified by the possibility of measuring it in routine standard exercise ECGs, usually affected by a high level of noise. Thus, aim of the present study was to evaluate the performances of 3 automatic TWA identification methods, namely the spectral (SM), the modified-moving-average (MMAM) and the adaptive-match-filter (AMFM) methods, on routine exercise ECGs with increasing heart rate (HR) of 46 implanted-cardiac-defibrillator patients (ICDP). TWA amplitude (TWAA) was measured in 1-min windows during the minimum (MinHR=83±12 bpm) and maximum (MaxHR=123±19 bpm) HR. At MinHR and MaxHR, the SM identified TWA in 12 and 23 ICDP, respectively (median TWAA: 4 μV and 29 μV, respectively; P<;10
^-3
), whereas the MMAM and AMFM identified TWA in all ICDP but the former method provided much higher median TWAA estimates (MinHR: 1507 μV, MaxHR: 1602 μV; P=0.5258) than the latter (MinHR: 22 μV, MaxHR: 38 μV; P<;10
^-4
). Eventually, only the SM and AMFM detected the expected significant TWAA increment at MaxHR. Thus, the AMFM appeared the most reliable method for TWA identification in routine exercise ECG recordings.