Session PB2.1

Detecting and Quantifying T-Wave Alternans in Patients with Heart Failure and Non-Ischemic Cardiomyopathy via Modified Spectral Method

A Ghaffari, MR Homaeinezhad, M Atarod, M Akraminia, M Davaeeha, B Hejrati*

K. N. Toosi University of Technology
Tehran, Iran

A new approach for the detection and quantitative classification of T-wave alternans (TWA) is developed in this study based on some modifications in conventional spectral method. To meet this end, the electrocardiogram (ECG) signals are first extracted from a Holter (MEDSET 3-channel chest electrodes, sampling frequency 1000 Hz, and 32 bits resolution) and ST/T segments are then detected using a robust ECG delineation algorithm. Afterwards, ST/T segments are aligned and using a new measure, motion of each segment is quantified relative to the previous and the next segments. In the next step, based on an appropriately-trained Multi-Class Binary Support Vector Machine, this measure is classified to "up-down", "all upward", ''all downward" and "indistinct" states of nature. Finally, the proposed algorithm is applied to the Holter data of patients with heart failure and non-ischemic cardiomyopathy, while the 12-lead Cambridge Heart-Bedford Massachusetts system, simultaneously showing the measured TWA, is attached to each patient. Up to now, values of 73.4% and 80.6% are achieved for Sensitivity and Positive Prediction, respectively.

(Abstract Control Number: 1)