Implantable antitachycardia devices (ICDs) have achieved overwhelming success in salvaging thousands of lives by providing immediate electrical therapy for the treatment of potentially lethal arrhythmias, i.e., ventricular tachycardia and ventricular fibrillation. These rhythms are believed responsible for over 80% of cases of sudden cardiac death, which claims 400,000 victims per year. The number of implants of ICDs is exceptional, over 100,000 implants to date, despite its relative infancy in the medical field. A large remaining problem to be solved in ICD technology is refinement of detection criteria such that the device no longer offers a simple brute force solution (if in question, shock). This is a three-fold problem: false shocks are an unnecessary patient distress; false shocks deplete battery power rendering the device less capable of addressing true urgencies and forcing premature explantation; and false shocks (or antitachycardia pacing) can initiate ventricular tachycardia (VT) or ventricular fibrillation (VF) when none previously existed. The specific aims of this proposal are: 1) to develop new signal processing techniques and pattern recognition schemes for accurate arrhythmia detection which exploit the present state of technology in ICDs; 2) to design improved diagnostic distinction between ventricular tachycardia and ventricular fibrillation which capitalize on the unique therapeutic choices for conversion of these arrhythmias; and 3) to refine methods of dual-chamber analysis for application to ICD arrhythmia diagnosis. Improved detection schemes comprise an engineering problem which must be addressed by modern digital signal processing techniques and imaginative engineering solutions employing concepts of lead-field theory and novel pattern recognition ideas. ***
