Single-photon emission computed tomography (SPECT) is playing an increasingly important role in modern medicine for imaging the brain, liver, kidneys, prostate, and other organs. The broad objectives of the project are to develop non-iterative methods for accurate and efficient reconstruction of three-dimensional (3D) images in SPECT and to evaluate these methods in clinical applications. Non-iterative methods for image reconstruction in SPECT are inherently computationally efficient. They are attractive in practice because they avoid problems that plague other methods. Most importantly, they can facilitate a closed-form: analysis of image statistics and aliasing, allowing the development of strategies for optimal suppression of the effects of noise, aliasing, and other errors. The research on non-iterative methods will significantly enhance the ability of SPECT to detect subtle lesions and to quantify accurately physiologic parameters in research and clinical applications. This proposal is intended specifically for further development of non-iterative, statistically optimal, computationally efficient, and numerically robust reconstruction methods in 3D SPECT and for evaluation of these methods in clinical SPECT studies such as In-111 Prostascint SPECT imaging of prostate cancer. We expect that the proposed methods will adequately compensate for the effects of photon attenuation, distance dependent spatial resolution, and data noise. We will apply methods developed by other investigators to compensate for scatter. We believe that our research will significantly strengthen the ability of SPECT for detecting subtle lesions in clinical applications.
The specific aims of the proposed research are (1) to fur ther develop non-iterative methods for optimal estimation of the ideal sinogram, (2) to develop adaptive and robust filtering approaches, (3) to investigate and mitigate the effects of additional sources of error, (4) to evaluate the proposed methods using phantom studies, and (5) to evaluate the proposed methods in clinical studies.
Showing the most recent 10 out of 15 publications
