AUTOMATED DETECTION OF ATRIAL FIBRILLATION DURING AN ELECTROPHYSIOLOGY STUDY

Abstract

The Electrophysiology (EP) study is performed by pacing from localized areas within the heart to detect different heart arrhythmia. On pacing, a patient, without any previous history, can develop “Atrial Fibrillation” (AF). EP specialists identify AF by visual monitoring of intra cardiac Electrograms (ICEMs) and experience. In order to proceed further with the diagnosis and treatment, the patient has to be taken out of AF. In this thesis a real time algorithm is developed for automatic detection of AF by analysis of ICEMs extracted from the High Right Atrium (HRA) catheter. This new algorithm continuously monitors ICEMs during a standard EP study. This work identifies two parameters, dominant frequency (DF) and Average power spectral ratio (APSR) from estimated power spectral density (PSD) of ICEMs. The ICEMs used are pre-processed for spectral estimation analysis. Non parametric power spectral estimation methods have been used detecting DF. APSR has been used to ensure the reliability of DF peak, and to differentiate between the atrial tachycardia, atrial flutter, and atrial fibrillation in their overlapping frequency ranges. The algorithm has been successfully tested on 2000 events, and the results have been validated by classified EP specialist.