Regions of interest in the venous sinuses as input functions for quantitative PET.

UNLABELLED As clinical PET becomes increasingly available, quantitative methods that are feasible in busy clinical settings are becoming necessary. We investigated the use of intracranial blood pools as sources of an input function for quantitative PET. METHODS We studied 25 patients after the intravenous injection of [18F]6-fluoro-L-m-tyrosine and compared sampled blood time-activity curves with those obtained in small regions of interest (ROIs) defined in the blood pools visible in the PET images. Because of the comparatively large dimensions of the blood pool at the confluence of the superior sagittal, straight and transverse sinuses, a venous ROI input function was chosen for further analysis. We applied simple corrections to the ROI-derived time-activity curves, deriving expressions for partial volume, spillover and partition of tracer between plasma and red blood cells. The results of graphic and compartmental analysis using both sampled [Cs(t)] and ROI [Cr(t)] venous input functions for each patient were compared. We also used an analytic approach to examine possible differences between venous and arterial input functions in the cerebral circulation. RESULTS Cr(t) peaked significantly earlier and higher than Cs(t) in this patient population, although the total integral under the curves did not differ significantly. We report some apparent differences in the results of modeling using the two input functions; however, neither the graphically determined influx constant, Ki, nor the model parameter that reflects presynaptic dopaminergic metabolism, k3, differed significantly between the two methods. The analytic results suggest that the venous ROI input function may be closer to the arterial supply of radiotracer to the brain than arterialized venous blood, at least in some patient populations. CONCLUSION We present a simple method of obtaining an input function for PET that is applicable to a wide range of tracers and quantitative methods and is feasible for diagnostic PET imaging.