Geometric model error reduction in inverse problem of electrocardiography

Date

2023-12-05

Author

Aldemir, Furkan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

163
views

188
downloads

Cite This

Electrocardiographic Imaging (ECGI) is a clinical tool for visualizing the electrical activity of the heart, detecting arrhythmias and mapping arrhythmic substrates. It has an inverse problem: obtaining electrical activity of the heart from body surface potential measurements, using a patient-specific model of the torso. This problem is an ill-posed one and does not have an exact solution. The ill-posed nature of the inverse problem could be handled with statistical constraints on the solution. These are based on prior statistical information about the solution. Several estimation methods exist to provide this regularization. This study focuses on the Bayesian MAP estimation. In most studies, MAP only handles the measurement noise and neglects the errors caused by discrepancies in the geometric modeling of the heart and the torso. This study, using measured electrogram data from the University of Utah, focuses on inspecting the effects of measurement noise and geometric model errors on the solution of the inverse problem. Further, some methods are proposed to compensate for these geometric model errors and improve the MAP solution. Methods are compared with respect to numeric metrics. There are also visualizations that compare these compensations, showing results on the heart map. Methods proposed in the thesis improve the performance of Bayesian MAP in estimating original epicardial potentials and activation times. Taking geometric model errors into account results in more accurate estimations of EP and pacing locations. This reduces the number of times the hearts of patients need to be modeled, which is an invasive procedure.

Subject Keywords

Electrocardiographic imaging, Inverse problem, Maximum a posteriori estimation (MAP), Geometric model errors

URI

https://hdl.handle.net/11511/107726

Collections

Graduate School of Natural and Applied Sciences, Thesis