The goal of the proposed research is to improve the acquisition, reconstruction, and extraction of quantitative information from Ga-67 SPECT data, and to assess these improvements to the imaging system using task-dependent criteria. Gallium has proven to be a useful nuclear medicine tracer for imaging certain tumors, and it is known that gallium avidity is correlated with histopathologic tumor grade. Imaging Ga-67, however, is challenging because it emits many high-energy photons. Quantitative estimates of Ga-67 tumor uptake are degraded by three principal sources of error: a location-dependent bias caused by imperfect correction for photon scatter and nonuniform attenuation, a size-dependent bias due to blurring by the nonstationary detector response function, and stochastic variability arising from Poisson noise in the acquired data. The proposed research will address these challenges by (1) optimizing for Ga-67 imaging several methods of correcting images for the effects of scatter and attenuation in the patient, (2) modifying for Ga-67 SPECT methods that have previously been developed for estimating activity within volumes of interest using a priori boundary information from registered CT images, and (3) designing a new collimator, tailored for Ga-67 quantitation in body imaging. These aspects of the imaging system will be optimized and evaluated on the basis of performance in several quantitative imaging tasks. The tasks to be considered, prototypes of tumor quantitation in the chest and abdomen, will involve estimation of activity concentration and size of lesions located in anatomically realistic backgrounds. For each of these tasks, Cramer-Rao lower bounds on variance or mean-squared error will be computed to determine the best possible performance for different correction methods, while maximum-likelihood or Bayesian parameter estimation will be used to measure best realized performance. Volume-of-interest activity estimation with resolution recovery will be used to assess clinically realizable performance. The investigators will measure, by simulation and phantom experiment, the improvements in performance in these tasks, and compare the results with theoretical bounds on performance. They will also consider clinical classification tasks related to non-Hodgkin s lymphoma. It is expect that the proposed imaging system improvements will lead to more accurate staging of lymphoma patients and, consequently, improved patient care due to enhanced capability to follow the progression of disease, choose the best treatment, and monitor the response to therapy.
