Filtering of the intraventricular electrogram and its effect on signal detection

Morphological waveform identification methods have been demonstrated to be effective for distinction of normal and abnormal conduction in intraventricular electrograms is small pilot tests. Yet these methods will find acceptance in future implantable devices only when computational demands and memory requirements are further reduced. A restricted passband allows reduced sampling rate and diminished storage for microprocessor based devices. In this study, the authors wished to determine the narrowest bandwidth that could be tolerated by a signal analysis morphometric (CWA) in order to distinguish ventricular tachycardia (VT) from sinus rhythm (SR). Twenty-five paired sets of bipolar intracardiac SR and monomorphic VT passages were analyzed. These signals were subjected to thirty-two different combinations of filter settings. Correlation coefficients were computed on a cycle-by-cycle basis for each signal passage to separate SR and VT. An absolute separation of ranges of CWA values was used as a discriminant function. The authors´ results showed that a high frequency cutoff of 80 Hz yielded adequate performance of morphological separation of normal and abnormal rhythm, given that low frequency cutoff was no higher than 10 Hz. Low pass filtering has minor impact on the performance of CWA compared to high pass filtering