System Identification applied to Cardiac Activation

Abstract: Clarifying the mechanisms maintaining atrial activity during atrial fibrillation (AF), still remains as a relevant topic. The purpose of this master thesis is to apply correlation analysis and system identification methods to study spatial and temporal propagation of atrial activation along coronary sinus (situated in the posterior left part of the heart, in the groove between left atrium and left ventricle) during paroxysmal atrial fibrillation (PAF) using data recorded catheter from 7 different patients. Furthermore, interatrial mechanisms of impulse conduction can be derived due to the position of coronary sinus. This study demonstrated consistency in electrical activity propagation during atrial fibrillation along coronary sinus in five patients out of six included. Nevertheless, direction and speed of propagation resulted dependent on the patient. The method was tried out during sinus rhythm (SR) obtaining the expected high consistency in propagation direction and speed which represented an interesting reference point to compare with atrial fibrillation results. Furthermore, linear relation among endocardial electrograms from coronary sinus and time invariant systems have been presented by computing simple linear models based on least squares method. Secondly, an impulse response method has been applied to reproduce atrial activations during atrial fibrillation and during sinus rhythm in ten endocardial electrograms from coronary sinus simultaneously. In this case, there is no need of measuring the input which is considered to be an impulse generated in the sinus node. This method was only useful in local prediction of atrial activations but not for prediction in the long term. Finally, Subspace Model Identification (SMI) methods have been applied to show an input-output relation among signals from inside the heart (endocardial electrograms from coronary sinus) and signals from outside the heart (V1 Surface-ECG).