Quantitative metabolic imaging of kinetic parameters from dynamic PET [F-18]-Fluorothymidine (FLT) data with application to studies in brain cancer patients

Objectives We consider direct non-parametric assessment of the FLT tissue residue or impulse response function. The method allows evaluation of tracer retention and delivery characteristics in contexts where there may be some aberrant pathology. The method can be viewed as an alternative to spectral mapping procedures based on generalized compartmental modeling.

Methods A statistical life-table formulation for the tissue tracer residue function is used. This has critical technical advantages over existing methods, including a guaranteed uniqueness of the target solution and simplified computational evaluation. Decomposition of the residue separates rapid vascular kinetics from slower blood-tissue exchanges and tissue retention. A mixture analysis scheme is used for application to voxel-level time-course data. The resulting kinetic mapping algorithm was applied to a series of 26 dynamic FLT brain tumor studies. Regions of interest (ROIs) corresponding to tumor and a number of normal brain structures were considered. Regional evaluation of kinetic parameters was carried out by direct averaging computed kinetic maps and also by application of standard 2-tissue compartment model to regional time course data extracted from raw dynamic FLT update images.

Results Simulation studies matched to cerebral FLT brain tumor studies characterize the reliability of regional quantitation of kinetic parameters (flux, flow and volume of distribution) as a function of dose, region size and regional heterogeneity. There is close agreement between (flux, flow and volume of distribution) parameters obtained by averaging over ROIs in computed parametric images and corresponding parameters derived by analysis of average ROI time-course data.

Conclusions The proposed kinetic mapping method is viable for routine processing of clinical FLT imaging in brain tumor studies.

Research Support Supported by National Cancer Institute under CA42045 and by Science Foundation Ireland under SFI PI-11/27.