SPATIO-TEMPORAL PATTERNS OF INTRACARDIAC SIGNAL FLUCTUATIONS DURING VENTRICULAR TACHYCARDIA IN MAN

The advent of multiple electrodes surface recording during the seventies has contributed importantly to improve our understanding of ventricular tachycardias (VT) mechanisms [1–7]. It has been also used to localise the arrhythmogenic substratum during open-chest surgery [8–11]. In the typical setting used at Sacre-Coeur Hospital [2], unipolar signals are collected by a set of electrodes positioned both on the epicardial surface (63 to 127 leads fixed to a sock pulled over the heart) and on the endocardial surface of the left ventricule (64 to 127 leads fixed on a balloon which is introduced through the mitral valve and inflated). During the surgery, long sequences of VT induced by burst pacing (driving at a fixed frequency, followed by a discharge of premature stimuli) are recorded. In the standard procedure, a few beats are selected for analysis. The activation times are determined for each electrode. They should correspond to the moments where an electrical front of activation travels through the tissue beneath the electrode. From these, isochronal maps are constructed describing the spatial propagation of the activation. In this analysis, many details of possible clinical relevance may be disregarded concerning the spatiotemporal evolution and the global features of the VT's. This paper presents some approaches to analyse globally protracted VT episodes. The first section describes some methods to speed up the determination of the activation times. The second section shows how principal component analysis [12] can be used to extract global features of the activation sequences.