The MRI-measured arterial input function resulting from a bolus injection of Gd-DTPA in a rat model of stroke slightly underestimates that of Gd-[14C]DTPA and marginally overestimates the blood-to-brain influx rate constant determined by Patlak plots.

The hypothesis that the arterial input function (AIF) of gadolinium-diethylenetriaminepentaacetic acid injected by intravenous bolus and measured by the change in the T(1)-relaxation rate (Delta R(1); R(1) = 1/T(1)) of superior sagittal sinus blood (AIF-I) approximates the AIF of (14)C-labeled gadolinium-diethylenetriaminepentaacetic acid measured in arterial blood (reference AIF) was tested in a rat stroke model (n = 13). Contrary to the hypothesis, the initial part of the Delta R(1)-time curve was underestimated, and the area under the normalized curve for AIF-I was about 15% lower than that for the reference AIF. Hypothetical AIFs for gadolinium-diethylenetriaminepentaacetic acid were derived from the reference AIF values and averaged to obtain a cohort-averaged AIF. Influx rate constants (K(i)) and proton distribution volumes at zero time (V(p) + V(o)) were estimated with Patlak plots of AIF-I, hypothetical AIFs, and cohort-averaged AIFs and tissue Delta R(1) data. For the regions of interest, the K(i)s estimated with AIF-I were slightly but not significantly higher than those obtained with hypothetical AIFs and cohort-averaged AIF. In contrast, V(p) + V(o) was significantly higher when calculated with AIF-I. Similar estimates of K(i) and V(p) + V(o) were obtained with hypothetical AIFs and cohort-averaged AIF. In summary, AIF-I underestimated the reference AIF; this shortcoming had little effect on the K(i) calculated by Patlak plot but produced a significant overestimation of V(p) + V(o).