Abstract
2321
Objectives The effect of preprocessing of blood data as well as the influence of some factors that may affect the fitting procedure were examined using the kinetic models to estimate the cerebral blood flow.
Methods O15-labeled butanol was administered intravenously to the healthy human as a bolus to estimate the regional CBF. The tissue time activity curve TAC was measured with the PET scanner and the TAC of the radioactivity concentration within the blood was obtained from a radial artery by a continuous measuring blood detector system. The influence of the correction of the input curve for dispersion and delay on the parameters K1 (flow) and PC (partition coefficient) was examined using the PKIN tool. Furthermore, different extent of smoothing of the corrected blood curve were tested. For each corrected input curve a 3 exponential-function was fitted to the outwash part of the curve. By using the corrected input curve, the TACs for mixed tissue (GM and WM) data was created to investigate the influence of the tissue heterogeneity.
Results It is found that the values of K1 and PC for mixed tissue in the 1-tissue compartment model and the flow and dispersion model were almost the same. The effect mainly appeared in the dispersion coefficient (DC) outcome in the latter model with maximum percentage error of -100% in an ROI of 100%, 60%, and 0% GM with smoothing value of 4. The effect of noise present in the tissue TACs on the fitting and the resulting CBF was also tested. It is found that K1, which were estimated by using dispersion uncorrected input curve were fluctuating in the model of flow and dispersion, and they were much more stable in all of selected GM-ROIs with already corrected input curve in the 1-tissue compartment model.
Conclusions The 3 exp fitting for the input curves gave approx. the same results of that came from unfitted curves therefore it is not necessary to consider the smoothing option and 3 exp function fitting for estimation of CBF. Furthermore it is recommended to correct the input curve before the fitting procedure takes place to obtain reliable results