A new method for fibrillation detection from endocardial electrograms

A new approach is presented for automatic efficient detection of atrial fibrillation not based on heart rate, but on morphologic analysis of the electrogram. Endocardiac signals have been collected from 9 patients (4 of them affected by chronic atrial fibrillation and 5 affected by other disturbances of the rhythm) during human cardiac electrophysiologic (EP) studies using a Manta Mennen Poligraph and cut into short intervals of different duration (from 0.5 to 20 seconds). 1018 seconds of sinus rhythm (SR) and 726 seconds of atrial fibrillation (AF) have been analyzed. Then the authors divided all segments in two separate groups: the training group to define appropriate discrimination threshold and the test group to quantify results. Results for the 4-seconds decision time show 94 and 96 values of 94 The method is straightforward and compatible with implantable devices (easy implementation and real time)