Accuracy of methods for calculating cerebral blood flow from intracarotid xenon-133 injection.

The accuracy of the three commonly used methods, the initial slope analysis, the stochastic analysis, and the compartmental analysis, for calculating mean cerebral blood flow from xenon-133 clearance curves was studied with the use of computer-generated and real curves. The accuracy of calculation was affected by the cutoff time of the curve, by the level of the compartmental blood flows to white and gray matter and by the ratio of these flow levels, by the relative weight of gray matter, and by the choice of the method of calculation. None of the methods was clearly superior to the others. Each had its own defects that render it more or less suitable for different situations. All three methods generally overestimated mean cerebral blood flow. This overestimation was greater the lower the flow. A curve-fitting index was devised which can be used to check the validity of the bicompartmental model when using compartmental analysis. This same index can provide, though not always, an estimate of the error in the calculation of mean cerebral blood flow when an optimization method is used.