Abstract
Based on data from routine intravenous Xe133-rCBF studies in 50 patients, using Obrist's algorithm the effect of counting rate statistics and amount of recirculating activity on reproducibility of results was investigated at five simulated counting rate levels. Dependence of the standard deviation of compartmental and noncompartmental flow parameters on recirculation and counting rate was determined by multiple linear regression analysis. Those regression equations permit determination of the optimum accuracy that may be expected from individual flow measurements. Mainly due to a delay of the start-of-fit time an exponential increase in standard deviation of flow measurements was observed as recirculation increased. At constant start-of-fit, however, a linear increase in standard deviation of compartmental flow parameters only was found, while noncompartmental results remained constant. Therefore, and in regard to other studies of potential sources of error, an upper limit of 2.5 min for the start-of-fit time and usage of noncompartmental flow parameters for measurements affected by high recirculation are suggested.