Reducing blood sampling requirements for quantitative PET: Some results for cerebral studies with FDG

The accurate recovery of detailed functional metabolic information from cerebral PET studies with FDG requires an estimate of the time course of the tracer in the arterial blood. We are developing an analysis scheme which would be useful for studies in which blood sampling is limited. The approach is based on the observation that, for a standardized injection protocol, subject-to-subject variability in arterial blood time activity curves is well described in terms of a low-dimensional parametric adjustment to a fixed template curve. Thus if an appropriate template is constructed, blood sampling may need be only sufficient to determine the parametric adjustment. A statistical technique for constructing such a template from archived data is developed. This technique is applied to a set of FDG blood curves. The results imply that the quantitation of PET data using blood curves constructed from an optimally adjusted template is similar to what would be obtained using the true blood curves. A method is presented for making the parametric adjustment to the template curve based on only limited blood samples. The median of error induced by the blood curve approximation scheme is about 4%. The results are encouraging.