Improved models for plasma radiometabolite correction and their impact on kinetic quantification in PET studies

The quantification of dynamic positron emission tomography studies performed with arterial sampling usually requires correcting the input function for the presence of radiometabolites by using a model of the plasma parent fraction (PPf). Here, we show how to include the duration of radioligand injection in the PPf model formulations to achieve a more physiologic description of the plasma measurements. This formulation (here called convoluted model) was tested on simulated data and on three datasets with different parent kinetics: [(11)C]NOP-1A, [(11)C]MePPEP, and [(11)C](R)-rolipram. Results showed that convoluted PPf models better described the fraction of unchanged parent in the plasma compared with standard models for all three datasets (weighted residuals sum of squares up to 25% lower). When considering the effect on tissue quantification, the overall impact on the total volume of distribution (VT) was low. However, the impact was significant and radioligand-dependent on the binding potential (BP) and the microparameters (K1, k2, k3, and k4). Simulated data confirmed that quantification is sensitive to different degrees to PPf model misspecification. Including the injection duration allows obtaining a more accurate correction of the input function for the presence of radiometabolites and this yields a more reliable quantification of the tissue parameters.