Adrian Lam

Ph.D. Defense Presentation

9:00 AM Tuesday, November 10th, 2015

Emory University Health Sciences Research Building (HSRB) E459

Advisor: 

John N. Oshinski, Ph.D., Department of Biomedical Engineering, Georgia Tech

Committee:

Xiaoping Hu, Ph.D., Department of Biomedical Engineering, Georgia Tech

Michael S. Lloyd, M.D., Division of Cardiology, Department of Medicine, Emory University School of Medicine

David Ku, Ph.D., Department of Radiology and Imaging Sciences, Emory University School of Medicine

Orlando Simonetti, Ph.D., Department of Biomedical Engineering, Ohio State University

Development of a Combined Angiography and Late Gadolinium Enhancement MR Sequence

With the rapid growth of catheter-based interventional cardiology procedures, knowledge of the vasculature surrounding the heart relative to the distribution of scar tissue can provide important information for procedural planning and appropriate patient selection. Two examples of this are: (1) selecting patients for Cardiac Resynchronization Therapy (CRT), or (2) evaluating the outcome for atrial fibrillation (AF) patients who have undergone Pulmonary Vein Isolation (PVI).

CRT uses a biventricular pacemaker to restore synchronous myocardial contraction in heart failure patients with evidence of ventricular dyssynchrony. Optimal improvement from therapy necessitates that the left ventricular pacing lead, which is transvenously implanted through the coronary veins, is situated at the latest contracting site that is not predominantly myocardial scar. In order to achieve this, co-registered images of the coronary veins and myocardial scar are necessary. However, current MR imaging protocols use separate sequences to image these features, complicating co-registration. 

PVI aims to circumferentially ablate around the pulmonary veins to electrically isolate triggers from reaching the left atrium. In order to evaluate ablation effectiveness, segmentation of the atrial wall is necessary. Yet, the thin nature of the atrial wall and low contrast at the blood-atrial wall interface make segmentation a challenge. While angiography images clearly delineate the inner atrial wall, co-registration is still necessary to the scar images.

These examples benefit from co-registration between MR angiography and scar images yet use different MR sequences making image registration cumbersome, preventing a unified display. The overall goal of this project was to develop a sequence that would produce inherently co-registered angiography and scar images using a single acquisition sequence and create image processing methods for a combined display. This sequence uses a slow infusion of gadolinium with a centric-ordered k-space acquisition scheme for angiographic imaging. Central k-space will be re-acquired at the end of the sequence (~10 mins after injection) and share outer k-space from the initial angiographic k-space to create inherently co-registered scar images. This sequence will be validated by imaging porcine hearts and then tested on two patient groups: (1) patients with known previous myocardial infarct (MI), and (2) patients with AF who have undergone PVI, to use LGE to evaluate ablation effectiveness. Algorithms will be developed to combine coronary vein and myocardial scar displays to allow LV lead planning. In addition, the concept of a pulmonary vein bullseye will be created to allow quantification of the extent of circumferential ablation in post-PVI patients.