Understanding Uncertainty and Improving Parameterization of In-Silico Cardiac Electrophysiology Models

Utah State Higher Education System--University Of Utah, Salt Lake City UT

Investigators

Abstract

7. Project Summary Atypical atrial flutter ablation outcomes are poor, with high recurrence rates following ablation procedures. Patient-specific computational models have recently shown promise as a tool for pre-ablation planning, with the ability to locate all possible reentrant circuits before a procedure. However, such models suffer from a lack of parameter optimization and high computational costs. The goal of this project is to apply state-of-the-art computational methods to improve the parameterization of patient-specific models of atypical atrial flutter. Aim 1 will investigate the effect changing input parameters has on the flutter circuits observed. By systematically varying these parameters, we will establish a quantitative relationship between model inputs and outputs. This will help us quantify the uncertainty in the models and identify the most critical parameters that influence the accuracy of the predictions. Aim 2 will look to increase the accuracy of scar detection from LGE-MRI using deep learning. Accurate detection of scar tissue from late gadolinium enhancement magnetic resonance imaging (LGE-MRI) is crucial for creating precise models. We will develop and test an overcomplete convolutional neural network (CNN) to improve scar segmentation, addressing challenges posed by image quality and scan variations. Aim 3 strives to personalize the electrophysiology parameters of the model. By tuning electrophysiology parameters to match patient-specific data, we will compare the accuracy of personalized versus generalized models in predicting clinically observed circuits. This personalization is expected to significantly enhance the predictive power of the models, making them more useful for clinical decision-making. Success in this project will improve the personalization of patient-specific atypical atrial flutter models, allowing for more accurate prediction and further clinical utilization. This project will also provide me with the technical expertise and scientific training required to be an independent research scientist in the field of cardiac electrophysiology and computing.