Calibration of a personalised model of left atrial electrophysiology
File(s)
Author(s)
Ali, Rheeda
Type
Thesis
Abstract
Patient-specific computer models of the human atria have the potential to aid clinical intervention
in the treatment of cardiac arrhythmias if suitably validated. An anatomically accurate,
patient-specific map of the electrical conductivity of the left atrial tissue is obtained using data
from delayed-enhancement magnetic resonance imaging (MRI), and validated against clinical
electroanatomic mapping measurements. The patient-specific intensity maps from two accepted
image segmentation and registration techniques are evaluated and compared, and the approach
is shown to be highly sensitive to the technique used. The segmentation technique and direction
of the maximum intensity projection are both critical in interpreting regions of fibrosis from
the patient specific intensity maps. The clinical data suggests a linear relationship between the
intensity and the local conduction velocity, which is incorporated into the atrial model through
calibration of the conductivity tensor. Simulations of the resulting atrial models produce activation
patterns that strongly correlate with clinical recordings. A novel semi-automated algorithm
is used for landmark selection for image registration for spatially comparing the electroanatomical
and MRI based images. The thesis concludes with a discussion of using the resulting computational
model to interrogate the underlying structural and functional substrates of patients
with atrial fibrillation.
in the treatment of cardiac arrhythmias if suitably validated. An anatomically accurate,
patient-specific map of the electrical conductivity of the left atrial tissue is obtained using data
from delayed-enhancement magnetic resonance imaging (MRI), and validated against clinical
electroanatomic mapping measurements. The patient-specific intensity maps from two accepted
image segmentation and registration techniques are evaluated and compared, and the approach
is shown to be highly sensitive to the technique used. The segmentation technique and direction
of the maximum intensity projection are both critical in interpreting regions of fibrosis from
the patient specific intensity maps. The clinical data suggests a linear relationship between the
intensity and the local conduction velocity, which is incorporated into the atrial model through
calibration of the conductivity tensor. Simulations of the resulting atrial models produce activation
patterns that strongly correlate with clinical recordings. A novel semi-automated algorithm
is used for landmark selection for image registration for spatially comparing the electroanatomical
and MRI based images. The thesis concludes with a discussion of using the resulting computational
model to interrogate the underlying structural and functional substrates of patients
with atrial fibrillation.
Version
Open Access
Date Issued
2016-08
Date Awarded
2016-12
Copyright Statement
Attribution NoDerivatives 4.0 International Licence (CC BY-ND)
Advisor
Tweedy, Jennifer
Peters, Nicholas
Sherwin, Spencer
Sponsor
British Heart Foundation
Publisher Department
Bioengineering, Aeronautics, NHLI
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)