Estimation of the Atrial Activity from Electrograms – a Beamforming Perspective

Tijs Moree1, Mathijs van Schie2, Natasja de Groot2, Richard Hendriks1
1Delft University of Technology, 2Erasmus University Medical Center


Aims: Inspection of the local atrial activity using atrial electrograms is important to determine the arrhythmogenic substrate underlying arrhythmias such as atrial fibrillation (AF). Using unipolar electrograms (EGM), the atrial component is clearly visible during sinus rhythm. However, during AF, the atrial activity is often distorted by ventricular far-field activity. An effective means to remove ventricular far-field activity is to use a bipolar electrode configuration. However, this configuration results in direction dependent distortions on the atrial components. From the field of array processing, the bipolar configuration is known as a differential beamformer. In addition to this beamformer, many other families of beamformers are known that provide better alternatives than the bipolar configuration, e.g., having both ventricular cancellation and distortionless properties on the atrial target signal.

Methods: To get more insight on the possibilities of estimating the atrial component from EGMs, we make use of concepts from the field of array processing. We formulate atrial component estimation as a beamforming problem and derive several atrial beamformers based on various signal model assumptions. The bipolar electrode configuration follows as a special case.

Results: Simulations are done with synthetically generated EGMs containing ventricular components and various degrees of AF. Compared to the bipolar configuration, the proposed estimators lead to significantly less distortion on the atrial component and a removal of the ventricular component. It also follows that with an increase of the degree of AF, the current signal model assumptions become somewhat less valid. Nevertheless, the performance of the proposed algorithms still significantly improves over the bipolar configuration.

Conclusion: The proposed algorithms led to a significant improvement over the use of the bipolar electrode configuration. For future work, the signal model could be further improved to even better handle higher degrees of AF.