Detectability of the epicardial breakthrough phenomenon from body surface potential maps by a linear predictive algorithm

The evaluation of tank-wall and epicardial potential field data of six isolated dog heart experiments was used to validate the simple linear predictive algorithm which was suggested earlier for the detection of characteristic events of ventricular activation (including epicardial breakthroughs). The validation of the algorithm was achieved by comparing the spatio-temporal estimated breakthrough parameters with those, obtained by the visual inspection of epicardial activation maps. Results suggest that most of the major epicardial breakthroughs can be detected accurately. In the six epicardial activation maps considered, 14 breakthroughs were identified. Eleven from these were correctly detected by the prediction error analysis, two were detected 2 msec (1 sample) earlier, one was missed while no false positive detection occurred