End-to-End Optimization of SPECT Instrumentation, Acquisition, and Reconstruction

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Single-Photon Emission Computed Tomography plays an important and well- validated role in the diagnosis and staging of a number of important cardiac diseases, and is widely used and available at many clinical sites in the US and around the world. The most common application has been myocardial perfusion SPECT to evaluate cardiovascular disease. A new and potentially important application is I-123 MIBG SPECT (AdreView) for myocardial innervation imaging. While there has been a great deal of work to develop improved reconstruction methods, and some work to develop optimized instrumentation and acquisition parameters, these have been done largely in isolation of each other. Questions such as: what is the optimal collimator when using collimator response compensation? what is the optimal energy window with scatter compensation?, is it optimal to use the same acquisition time for each projection view?, and is it optimal to use the same acquisition time per patient? have never been addressed. In addition, there are currently a number of dedicated scanners for cardiac imaging that incorporate several novel features to decrease acquisition time. Some of these are based on cadmium zinc telluride semiconductor detectors that provide improved full-width at half maximum energy resolution, but have a more complicated energy response that includes tails extending to low-energies. Other new detector materials such as LaBr with improved energy resolutions are on the horizon. However, energy windows and scatter compensation methods for CZT and LaBr detectors have not been optimized, nor has their benefit on cardiac imaging been rigorously tested. In this grant we propose to use novel, state-of-the art task-based image quality measures and realistic well-validated simulations calibrated by clinical studies to perform comprehensive end-to-end optimization of instrumentation, acquisition, reconstruction, and compensation parameters and methods. We will investigate the tradeoff between these factors and acquisition time/injected dose, allowing physicians to trade image quality for reduced radiation dose. The results would have an immediate impact in providing improved image quality, diagnostic accuracy, and reduced patient dose, while also having a longer term impact by guiding the development of future SPECT systems. The work will also provide validation of novel optimization strategies using projection-domain ideal-observers that handle mismatch between the imaging model used in the reconstruction and the true imaging process.