Session S92.5

A New Method for Atrial Electrical Activity Analysis from Surface ECG Signals Using an Energy Ratio Measure

N Weissman*, A Katz, Y Zigel

Ben Gurion University of the Negev
Be'er Sheva, Israel

Analysis of atrial activity (AA) can be a great help for the physician who tries to identify the arrhythmia from ECG signals. In many cases, the P wave is hidden in QRS complex or in T wave, and cannot be observed. We present a new method for the analysis of AA using a linear combination of the 12 lead ECG recording.
It has been shown that independent component analysis (ICA) can separate the AA from the ECG recordings in the atrial fibrillation (AF) case. ICA relies on three assumptions: The AA can be constructed by a linear combination of the 12 lead ECG, the independence of the sources, and the sub-Gaussianity of the AA; these assumptions stand in the AF case, but not necessary in other arrhythmias such as AV nodal reentrant tachycardia (AVNRT). Our proposed method, the energy ratio based linear combination (ERBLC), relies only on the first assumption and on the physician, who needs to identify and mark only one atrial electrical activity signal (one segment) in the surface ECG. The method forces the linear combination to converge to the AA source signal by creating a cost function that is the energy ratio between the marked segment and the rest of the signal. Using the gradient ascent optimization algorithm, the linear combination of the 12 lead ECG recordings converge to one signal that produce the maximum of the cost function, and hence, emphasizes the atrial activity in the signal.
We used 10 8-second segments from 12-lead ECG signals which include several cases of AF, AVNRT and sinus rhythm. We used intra-cardiac recordings for comparing the results and performance evaluation. In the AF case, the proposed ERBLC method performance was similar to the ICA method. In the AVNRT and the sinus rhythm cases, the ERBLC performance was much better than the ICA method, nevertheless the ventricular activity (VA) was drastically reduced but has not been cancelled. The conclusions are that the proposed ERBLC method can emphasize the AA in the surface ECG signals, and can be a valuable tool for the physician in identifying and analyzing cardiac arrhythmias.

(Abstract Control Number: 57)