Abstract
Myocardial infarction can cause ventricular tachycardia as a result of reentrant electrical activation waves propagating around the infarct scar. The tachycardia can be treated by radiofrequency catheter ablation which requires the cardiologist to deliver radiofrequency, via an intracardiac catheter, to ablate a specific site within the scar, disrupting the reentrant circuit, to terminate the arrhythmia. Therefore, determining the location of the scar is an important step in the procedure. MRI and CT scans can show the region of scar but are costly and are contraindicated in many cases. Cardiologists observe ECG recordings of the patient's tachycardia, looking at the gross characteristics of the signal to determine an approximate location of the scar. However, this technique assumes a recording of the patient's tachycardia is available, and is only able to suggest a gross region of the heart. In addition, the method is based on studies which have identified characteristic features of the ECG using intracardiac mapping to determine the location of the scar, which may be unreliable.
This study aims to determine features of the ECG which may be able to predict the location of an infarct scar with more accuracy and specificity than current methods allow. The use of computational models ensures that the true location of the scar is known, unlike in previous studies. Moreover, we aim to determine whether there are any characteristics of resting state ECG which may indicate the location of an infarct scar.
An anatomically accurate finite element model of rabbit ventricles in a conductive bath was utilised in order to simulate electrical activation waves and generate ECG signals, by reconstructing the extracellular potentials. Scar regions comprising of non-conducting scar surrounded by tissue with altered electrophysiological properties to represent the borderzone were incorporated into the ventricular model at varying locations across the myocardium. The models were stimulated using an S1S2 protocol to produce wave block and reentry. ECGs were reconstructed and the differences between models were observed.
Results suggest that differences in timing and amplitude of the R wave on the ECG could be an indication of scar location. Changes in the repolarisation phase of the ECG were also apparent, suggesting more features which could determine the location of the scar. Importantly, characteristic features of the ECG could also be determined from resting state ECG, generated from models where scar was present but no reentry occurred.
Utilising computational models of rabbit ventricles with scars incorporated at a variety of locations around the myocardium, we were able to determine a set of features from the ECG which may be of use in determining the location of an infarct scar. Future validation of this study using patient data could indicate that this methodology may be of use in predicting scar location in ablation procedures.
Acknowledgement/Funding
YH is funded by the MRC (MR/R024995/1). JT acknowledges the financial support of the EPSRC (EP/N014391/1) and the Wellcome Trust WT105618MA.
Computational Representations of Myocardial Infarct Scars and Implications for Arrhythmogenesis
