Computer Discrimination of Atrial Fibrillation and Regular Atrial Rhythms from Intra-Atrial Electrograms JANET SLOCUM,* ALAN SAHAKIAN, and STEVEN SWIRYN Department of Electrical Engineering, Northwestern University and the Division of Cardiology, Department of Medicine, Evanston Hospital; and Northwestern University Medical School, Evanston. IL SLOCUM, J., ET AL.: Computer discrimination of atrial fibrillation and regular atrial rhythms from intra-atrial electrograms. Heliahle detection of atrial fihriilation from intra-atrial data is an important requirement for automatic impJantable anti-tachycardia devices. Simultaneous filtered and unfiltered intra-atrial electrograms were recorded from patients in regular rhythms (12 sinus rhythms and six regular atrial tachycardias) and atrial fibriJiation fnine rhythms). Each rhythm was broken down into consecutive 4-second data segments for analysis by atrial rate caicuiation, power spectrum analysis and amplitude probability density function generation. Significant diferences were found between regular rhythms and atriaJ fibrillation for atrial rate, for the percentage of the total power in the 4-9 hertz band and for amplitude probability density close to the isoelectric region. There was no overlap for any of these three parameters. For each method of analysis, algorithms were generated to discriminate individual data segments from regular rhythms and atriai fibrillation with high sensitivity and specificity. Comparable results were found when sinus rhythm was excluded from the analysis. Characteristics of intra-atrial recordings during atrial fibrillation were remarkably similar to previously published reports of intra-ven- tricular recordings during ventricular ^briilation. Each of the three n\ethods of analysis may provide an algorithm for accurate detection of atrial fibrillation by anti-tachycardia devices. (PACE, Vol 11 May 1988] intracardiac recordings, antitachycardia device, electrograms, spectrum, atriaJ fibrillation, arrhythmia Introduction The recent development of implantable anti- tachycardia devices has created a need for an al- gorithm to recognize atrial fibrillation from intra- atrial electrograms.^ This need reflects the possihility that atrial fihriilation may he inadvertently initiated during attempted conversion of other rhythms,^"" that devices are currently heing developed for auto- matic termination of atrialfibrillation,^"^and that * Present address: Mortara Instruments, 7865 North 86th Street, Milwaukee. WI 53224. Address for reprints: Steven Swiryn MD, Cardiac Electrophys- iology, Rm 300A. Burch Hall, Evanston Hospital, 2650 Ridge Ave, Evanston, IL 60201. Received May 20, 1987; accepted July 6. 1987. patients treated with antitachycardia devices are often at risk for spontaneous episodes of atrial fi- brillation. The increased ventricular rate result- ing from atrial fibrillation may be detected by current devices and initiate inappropriate ther- apy.''•^"^° There have been few previous reports characterizing intra-atrial electrograms during atrial fibrillation."'^^ We have therefore tested several methods of analysis, including rate, to differentiate atrial fihriilation and regular atrial rhythms from intra-atrial recordings. Methods Bipolar intra-atrial electrograms (with a 1 cm inter-electrode distance)** from patients in sinus " USCI, Billerica, MA, USA. 610 May 1988 PACE, Vol. 11