Variability study of a non-invasive approach to the absolute quantification of cerebral blood flow with Tc-99(m)-ECD using aortic activity as the arterial input estimate

The absolute measurement of regional cerebral blood flow can be made non-in vasively if dynamic scanning is performed immediately following the intrave nous injection of Tc-99(m)-labelled HMPAO or ECD. By application of an elem entary one-step kinetic model in which the arterial input curve is estimate d from a region of interest over the anterior aortic arch projection, a bra in perfusion index (BPI) can be calculated which is correlated to the total hemispheric blood flow. The aim of this study was to assess the variabilit y of the BPI approach as applied to the calculation of cerebral blood flow For 46 patients (30 men, 16 women) aged 23.7-74.4 years (mean 60.3), we cal culated hemispheric BPI factors. For two subgroups of 10 patients chosen at random, several data-processing parameters, including the number of noise suppression steps and the linear slope interval for the Patlak fit, were as sessed. To determine the influence of region-of-interest selection and to e stimate intra- and inter-observer differences, three independent observers conducted a repeat analysis of all 46 patients. From five patients who unde rwent the same injection procedure twice, an estimate of intra-subject repr oducibility was calculated. Temporal noise reduction of the input data sign ificantly decreased the intra-observer variability. Visual estimation of th e linear interval in the Patlak curve is much less reproducible than automa tic interval selection; very high interval borders in particular significan tly decrease the calculated BPI. The intra-observer coefficients of variati on for the BPI were 5.3, 5.6 and 7.9% respectively, corresponding to 2.8, 3 .0 and 4.1% when converted to hemispheric CBE The slight inter-observer dif ferences found could be attributed to differing processing parameters. For the five patients in whom the procedure was repeated, the hemispheric BPI r eproducibility index was on average 2.6%. Determination of absolute hemisph eric blood flow can be performed in a precise manner by means of a dynamic scan during the first 2 min after tracer injection after careful optimizati on of the processing parameters. ((C) 1999 Lippincott Williams & Wilkins).