Our general goal is automatic identification of arrhythmias within implantable devices by computer-based signal processing of the intracardiac electrograms. This goal appears to be realizable because: 1) the arrhythmias to be identified can be provoked in the catheter electrophysiology laboratory and the resulting electrograms can be recorded for detailed analysis; 2) modern signal processing techniques exist for characterizing the electrograms in the time domain and by spectral content; 3) computer algorithms can be designed to recognize these characteristic electrogram features in each arrhythmia; 4) implantable microcomputers are available in which the algorithms can be implemented. An important premise of this proposal is that the hardware for electrical termination of arrhythmia is already availbale and thus there is a risk of new generations of devices being implanted for arrhythmia control before the underlying electrophysiologic concepts and pattern recognition rules for identifying the arrhythmia are developed and tested. The signal processing algorithms and decision rules for automatic detection and discrimination will be implemented in real time in a computer program small enough to be part of an implanted anti-tachycardia or defibrillating device. We will first examine and describe in both the time and frequency domains the characteristic features of electrograms recorded during provocative electrophysiologic study in animals and humans. Then we will develop computer algorithms for signal processing of the electrograms and for arrhythmia pattern recognition. Finally, we will test the stated hypotheses by conducting additional animal and human electrophysiology studies and determining the sensitivity and specificity of the algorithms. Our approach combines the engineering skill of computer-based signal analysis with the medical skill of diagnostic cardiac electrophysiology. It is our expectation that this approach will provide a sound scientific basis upon which development of specific hardware by pacemaker manufacturers can proceed.
Showing the most recent 10 out of 14 publications
