ESTIMATION ALGORITHMS FOR DYNAMIC TRACER STUDIES USING POSITRON-EMISSION TOMOGRAPHY.

Algorithms are developed for estimating statistics for use by parameter estimation algorithms in dynamic tracer studies utilizing positron-emission tomography and requiring high temporal resolution. Two types of statistics are considered. One can be used with the expectation-maximization algorithm to compute maximum likelihood parameter estimates, and the other computes a histogram of activity levels versus time for use with weighted least-squares parameter estimation algorithms. An estimator of the variance of this histogram is given. Variants for use with both time-of-flight and projection data collected at high frame rates are presented. The algorithms account for the effects of attenuation, randoms, detector efficiency, and nonuniform sampling.