The goal of the proposed research is to improve the accuracy in identifying patients with coronary artery disease by single photon emission computed tomographic (SPECT) myocardial perfusion imaging. To attain this goal the investigations will complete the development of the corrections for the degradations inherent in imaging (non-uniform attenuation, scatter, distance-dependent resolution (DDR), physiological motion, and noise) begun during the original grant period.
The specific aims are: 1) to investigate the limits imposed by variations in cardiac wall thickness, motion, and contractility, and to determine to what extent alterations in acquisition and reconstruction strategies can overcome the current limitations; 2) to investigate the impact of respiratory motion on defect detection and develop strategies to diminish this impact; 3) to investigate methods to reduce the influence of noise in reconstructed static (3D) and gated (4D) slices; and 4) to conduct two receiver operating characteristics (ROC) studies using clinical images to determine the relative detection accuracy of the acquisition and reconstruction strategies developed herein. The investigations of the first three specific aims will use simulations based on anatomical models of the left ventricle (LV) derived from the segmentation and fitting of gated, breath-held magnetic resonance imaging (gMRI) slices from normal subjects. By utilizing simulation studies we can systematically evaluate acquisition and reconstruction strategies which may not yet be clinically feasible. The comparison criteria used in the simulation studies will include the degree of uniformity of maximal-count circumferential-profile polar-maps, and defect detection by numerical-observers, human-observers, and quantitative analysis. The first clinical human-observer ROC study will compare stress images reconstructed by filtered backprojection (FBP) with no compensation, and those rendered using iterative reconstruction with compensation for: 1) attenuation; 2) attenuation and scatter; and 3) attenuation, scatter and DDR. The second ROC study will compare: 1) iterative reconstruction with the best combination of compensations, as determined in the first ROC study, versus FBP reconstruction with no compensation; 2) expert readers (board certified physicians) versus readers with more limited expertise (cardiology fellows); and 3) reading only the stress slices versus using all of the scintigraphic information which is routinely clinically available. Overall, the proposed investigations should permit us to develop an optimal imaging strategy for myocardial perfusion imaging, and determine whether this strategy enhances clinical interpretation for readers with differing skill levels.
Showing the most recent 10 out of 27 publications
