Non-invasive dynamic tracking of cardiac vulnerability by analysis of

Abstract

A non-invasive method for dynamic tracking of cardiac vulnerability to ventricular fibrillation is disclosed. A heart is monitored to sense an ECG signal. The sensed ECG signal is then amplified and low-pass filtered before it is digitally sampled and stored. The location of the T-wave in each R-R interval (pulse) of the ECG is estimated and each T-wave is partitioned into a plurality of time divisions. The sampled ECG signal in each of the time divisions is summed together and a time series is formed for each of the time divisions such that each time series includes corresponding sums from corresponding time divisions from successive T-waves. Each time series is detrended in order to eliminate the effects of drift and DC bias, and then a method of dynamic estimation is performed on each time series to estimate the amplitude of alternation for each time division. The methods of dynamic estimation include Complex Demodulation, Estimation by Subtraction, Least Squares Estimation, Auto Regressive Estimation, and Auto Regressive Moving Average Estimation. In this manner, the cycle-to-cycle alternation in the T-wave of the ECG can be determined for dynamic tracking of cardiac vulnerability. In one embodiment of the invention, the ECG is sensed non-invasively via the precordial or chest leads. In alternate embodiments, the ECG is sensed via a catheter inserted into the apex of either the left or right ventricles of the heart.