A mentored training program in quantitative medical imaging

点击翻译按钮获取中文摘要

DESCRIPTION (provided by candidate): Single Photon Emission Computed Tomography (SPECT) is a popular nuclear medicine technique in clinical practice and medical research. In SPECT, a labeled radionuclide is introduced into a subject, via injection, for example. The emitted photons are then collected/counted by rotating gamma cameras. At each position the camera collects a two dimensional projection of the organ under study. This situation is rare from a statistical viewpoint because the Poisson distribution of the decay of the emitted photons is known. The challenge is to estimate the mean rate of this distribution when only observing the projection data. The primary goal of this grant is to motivate and develop a rigorous training program in the physics and instrumentation of medical imaging reconstruction focusing on the SPECT reconstruction problem. This application, proposes a didactic and research training program to build a foundation in the science of medical imaging. The specific aims of this proposal build on the success of iterative algorithms for SPECT image reconstruction. Despite this success, current practices suffer from long run times and ad hoc procedures for stopping the algorithm. The aims of this proposal are to accelerate EM based iterative reconstruction algorithms and to theoretically and empirically investigate intra-iteration smoothing. All of the developed algorithms will be extensively tested using Monte Carlo and actual patient data.