Date of Award

Summer 2020

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

School

Computing Sciences and Computer Engineering

Committee Chair

Chaoyang Zhang

Committee Chair School

Computing Sciences and Computer Engineering

Committee Member 2

Weihua Zhou

Committee Member 3

Huiqing Zhu

Committee Member 3 School

Mathematics and Natural Sciences

Committee Member 4

Bikramjit Banerjee

Committee Member 4 School

Computing Sciences and Computer Engineering

Committee Member 5

Robert Bober

Committee Member 6

Zhaoxian Zhou

Committee Member 6 School

Computing Sciences and Computer Engineering

Abstract

OBJECTIVES. Coronary artery disease (CAD) is the most common type of heart disease and kills over 360,000 people a year in the United States. Myocardial revascularization (MR) is a standard interventional treatment for patients with stable CAD. Fluoroscopy angiography is real-time anatomical imaging and routinely used to guide MR by visually estimating the percent stenosis of coronary arteries. However, a lot of patients do not benefit from the anatomical information-guided MR without functional testing. Single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) is a widely used functional testing for CAD evaluation but limits to the absence of anatomical information. Therefore, the integration of SPECT-MPI and fluoroscopy angiography becomes necessary to guide MR. This study aims to develop a 3D image-guided system that integrates fluoroscopy angiography with SPECT-MPI to improve MR decision-making.

METHODS. Major coronary arteries on fluoroscopy angiograms were extracted with a deep learning-based image segmentation model and then reconstructed into 3D artery anatomy using computer stereo geometry calibration and reconstruction algorithms. 3D LV epicardial surface was extracted from SPECT-MPI using a dynamic programming-based algorithm. Artery-surface fusion was completed with non-rigid registration based on a scaling iterative closest points (S-ICP) algorithm and post-processing techniques based on a line-surface overlay algorithm. The accuracy of artery-surface fusion was evaluated via both computer simulation and real patient data. For technical evaluation, simulated fluoroscopy angiograms and nuclear myocardial perfusion images were integrated using the reconstruction and fusion approaches and then validated against the digital X-CAT phantom. For clinical validation, computed tomography (CT) angiograms were used as the ground truth: LV & right ventricular (RV) epicardial surfaces and arteries were manually extracted on CT angiograms and then registered with the SPECT epicardial surfaces; the locations of fluoroscopic and CT arteries on the SPECT epicardial surface were compared and evaluated.

RESULTS. In the technical evaluation, the distance-based mismatch error between simulated fluoroscopy and phantom arteries is 1.86±1.43 mm for left coronary arteries (LCA) and 2.21±2.50 mm for right coronary arteries (RCA). In the clinical validation, thirty patients with coronary artery disease (CAD) were enrolled. The mean distances between the fluoroscopic and CT arteries were 3.84 ± 3.15 mm for LCA and 5.55 ± 3.64 mm for RCA. The concordance between the fluoroscopic and CT arteries in the clinical 17-segment model agreed well with a Kappa value of 0.91 (95% confidence interval (CI): 0.89 to 0.93) for LCA and a Kappa value of 0.80 (95%CI: 0.67 to 0.92) for RCA.

CONCLUSION. A 3D image-guided system for MR decision-making has been developed that reconstructs 3D arterial anatomy from fluoroscopy angiograms, extracts LV epicardial surface from SPECT myocardial perfusion images, and fuses the arterial anatomy with LV epicardial surface. The system is technically accurate to guide MR decision-making and clinically feasible to be used in the catheterization laboratory. With the precise integration of anatomical and physiological information, it has a great promise to improve the decision-making and outcome of MR in clinic practice.

Share

COinS