A new detection algorithm for implantable cardioverter defibrillator

One of the most important subsystems of implantable cardioverter defibrilla tor (ICD) is the sensing stage, since it determines the sensitivity and spe cificity of the device to detect the heart rate and the underlying arrhythm ia. This paper aims to investigate a new detection algorithm for ICD, which operates fully automatically. The algorithm ARGUS was implemented as a com puter model and tested with intracardiac electrograms recorded (band-pass: 0.05 to 500 Hz; sampling rate: 1-4 kHz) under different rhythm condition li ke sinus rhythm (n = 18), atrial tachycardia (n = 16), and ventricular tach ycardia as well as fibrillation (n = 139) during electrophysiological tests or ICD implantation. The results of the tests were visually inspected on a beat-to-beat basis. In total 31934 events were classified by the algorithm (18758 as long intervals (LI) with cycle length > 300 ms; 13176 as short i ntervals (SI)). 195 out of the 13176 SI and 572 out of 18758 LI were incorr ectly classified (SI: 1.48%; LI: 3.05%). In conclusion the new algorithm yi eld high sensitivity (99.9%) and specificity (97.0%) as known from conventi onal ICD algorithms but need no manual adjustments.