ROC EVALUATION OF TC-99M CARDIAC SPECT METHODS

点击翻译按钮获取中文摘要

DESCRIPTION (Adapted from Applicant's Abstract): The applicants have proposed to improve the image quality and diagnostic effectiveness of Tc-99m myocardial perfusion imaging with single photon emission computed tomography (SPECT). The importance of this clinical procedure in the diagnosis of heart disease is clearly established, and there is a need for the development and careful evaluation of improved imaging methods. This application will address several issues in Tc-99m myocardial SPECT, presenting new methods for improving the quality of the SPECT image and evaluating the methods using receiver operating characteristic (ROC) analysis. The ROC studies will use data acquired from anthropomorphic thorax phantoms that model the attenuation and source distributions of patients during Tc-99m myocardial perfusion imaging. The ability of human observers to detect simulated perfusion defects will be measured in the reconstructed phantom image obtained using the proposed SPECT methods. The specific issues that will be considered include: 1.) 360 versus 180 degree data acquisition and reconstruction. Image quality can be greatly affected by this choice, and for defect detection in Tc-99m myocardial perfusion SPECT with multihead cameras, it is not clear which technique is superior. These alternative approaches will be investigated, considering the effect of defect location and body size; 2.) Non-uniform attenuation compensation using acquire transmission data. Acquisition and processing methods for transmission computed tomography (TCT) with a three-headed SPECT system will be presented and evaluated. The TCT system design used a Tc-99m line source with fan beam collimator. The acquisition is complicated by the presence of the Tc-99m perfusion tracer within the body, and methods are proposed for reducing the influence of the tracer in the TCT data. Non-uniform attenuation compensation of the SPECT image using the reconstructed TCT data will be compared with uniform and no compensation; and 3.) Three- dimensional iterative reconstruction. Iterative reconstruction methods will be developed and evaluated that model in three dimensions the important physical processes in SPECT imaging. The effects of attenuation, scatter and detector response will be incorporated into iterative algorithms, such as maximum likelihood-expectation maximization (ML-EM), and compared with the filtered backprojection (FB) methods considered earlier in the proposed study.