Tachycardia and fibrillation detection by automatic implantable cardioverter-defibrillators: sequential testing in time domain

The automatic implantable cardioverter-defibrillator (AICD), a device that delivers a high-energy shock (10-30 J) directly to the heart when life-threatening arrhythmias are detected, must accurately detect ventricular fibrillation (VF) and ventricular tachycardia (VT) rhythms. The signal-processing problem is how to accurately distinguish normal sinus rhythm or supraventricular tachycardia from malignant VT and VF signals. An approach based on discriminating probability distributions of interbeat intervals in electrogram signals recorded from atrial and ventricular leads is described. A sequential hypothesis-testing algorithm that allows desired false positive and false negative rates and the time it takes to detect the arrhythmias to be preset is presented. A tradeoff between reliability and duration of test is noted; a decision can be reached in a shorter time if higher error rates are accepted as permissible. In an analysis of 85 cases of VF and 85 cases of VT, 53% of VF cases and 66% of VT cases were classified after 1 s, 97.6% of VF cases and 100% of VT cases were classified after 5 s, and within 7 s, all the remaining VF cases were classified.