Innovative reconstruction algorithms for undersampled SPECT

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): This project will develop an innovative reconstruction approach for single-photon emission computed tomography (SPECT) based on compressed sensing theory. The goal of the proposed research is to enable dynamic images of tracer uptake (~ 1 second per image) with a minimal number of cameras, as cameras represent the major cost of a SPECT system. Reducing the number of cameras increases the likelihood that dynamic small-animal SPECT systems can be successfully commercialized and made accessible to the general research community. The kinetic parameters that can be estimated from dynamic studies provide important information about physiological mechanisms and disease states. Compressed sensing algorithms have shown promise for reconstructing undersampled data from Magnetic Resonance (MR) and Computed Tomography (CT) acquisitions. The proposed project will develop a novel optimization-based reconstruction algorithm for SPECT imaging, drawing on compressed sensing theory. Unlike previously proposed algorithms for MR and CT, our proposed algorithm will model the Poisson noise statistics of SPECT and use spline wavelets as a novel sparsifying transform. The proposed CS algorithm will be studied through simulations, phantom experiments, and animal data obtained from the three-camera small-animal SPECT system at the Pulmonary Physiology and Research Laboratory. The performance of the algorithm will be quantified and compared to conventional reconstruction approaches with respect to quantitative metrics of spatial and temporal accuracy. The algorithm will be tested with respect to the expected clinical task of imaging rapid (~ 1 second) tracer uptake through simulations, dynamic phantom experiments, and animal data. This project is in collaboration with the Pulmonary Physiology and Research Laboratory at the Zablocki VA Medical Center, where the kinetic information from the reconstructed time-activity curves will be used to develop tracers for diagnosing lung injury and disease. The successful completion of this project will result in a novel reconstruction approach that will benefit numerous SPECT applications, for example cardiac and oncology studies. PUBLIC HEALTH RELEVANCE: This project will develop innovative reconstruction algorithms for rapid (~ 1 sec), dynamic single-photon emission computed tomography (SPECT) acquisitions. Successful completion of the project will result in algorithms for constructing images of the dynamic tracer uptake, which will facilitate estimation of the underlying kinetic parameters. In our laboratory, the algorithms will be used to develop imaging agents for diagnosing the extent of lung injury. The ability to accurately quantify the fast (~1 sec) 3D tracer uptake in vivo will be beneficial for numerous small-animal and clinical studies, for example in the areas cancer imaging.