Resumen de Multimodal ventricular tachycardia analysis: towards the accurate parametrization of predictive hpc electrophysiological computational models
After a myocardial infarction, the affected areas of the cardiac tissue suffer changes in their electrical and mechanical properties. This post-infarction scar tissue has been related with a particular type of arrhythmia: ventricular tachycardia (VT). A thorough study on the experimental data acquired with clinical tools is presented in this thesis with the objective of defining the limitations of the clinical data towards predictive computational models. Computational models have a large potential as predictive tools for VT, but the verification, validation and uncertain quantification of the numerical results is required before they can be employed as a clinical tool.
Swine experimental data from an invasive electrophysiological study and Cardiac Magnetic Resonance imaging is processed to obtain accurate characterizations of the post-infarction scar. Based on the results, the limitation of each technique is described. Furthermore, the volume of the scar is evaluated as marker for post-infarction VT induction mechanisms.
A control case from the animal experimental protocol is employed to build a simulation scenario in which biventricular simulations are done using a detailed cell model adapted to the ionic currents present in the swine myocytes. The uncertainty of the model derived from diffusion and fibre orientation is quantified. Finally, the recovery of the model to an extrastimulus is compared to experimental data by computationally reproducing an S1-S2 protocol.
Results from the cardiac computational model show that the propagation wave patterns from numerical results match the one described by the experimental activation maps if the DTI fibre orientations are used. The electrophysiological activation is sensitive to fibre orientation. Therefore simulations including the fibre orientations from DTI are able to reproduce a physiological wave propagation pattern. The diffusion coefficients highly determine the conduction velocity. The S1-S2 protocol produced restitution curves that have similar slopes to the experimental curves.
This work is a first step forward towards validation of cardiac electrophysiology simulations. Future work will address the limitations about optimal parametrization of the O'Hara-Rudy cell model to fully validate the cardiac computational model for prediction of VT inducibility.
