Simultaneous estimation of a model-derived input function for quantifying cerebral glucose metabolism with [18F]FDG PET

Abstract

Background: Quantification of cerebral glucose metabolism by dynamic [ 18 F]FDG PET is often limited to the cerebral metabolic rate of glucose (CMRGlu) due to the challenges associated with obtaining microparameters at the voxelwise level (e.g., noise, choice of parameter estimation method, and input function quality). Efforts to avoid invasive arterial sampling include the simultaneous estimation (SIME) approach, which models the image-derived input function (IDIF) by series of exponentials with coefficients obtained by fitting time activity curves (TACs) from multiple volumes-of-interest. A limitation of SIME is the assumption that the input function can be modelled accurately by a series of exponentials. Alternatively, we propose a SIME approach based on the two-tissue compartment model to extract a high signal-to-noise ratio (SNR) modelderived input function (MDIF) from the whole-brain TAC. The purpose of this study is to present the MDIF approach and its implementation in the analysis of animal and human data. Methods: Simulations were performed to assess the MDIF approach in terms of accuracy, by evaluating a range of temporal resolutions; and precision, by adding noise to simulated TACs. The derived MDIFs were compared to the gold standard using retrospective data from animal experiments ( n = 5). Retrospective data from neurologically healthy volunteers ( n = 18) were used to extract macro- and microparameters from dynamic [ 18 F]FDG PET data. Results: Simulations demonstrated that extracting the MDIF from a whole-brain TAC requires using a sampling rate sufficient to minimize truncation errors. The MDIF approach improved the precision of estimated [ 18 F]FDG microparameters compared to the original SIME method. Good agreement between MDIFs and measured AIFs was found in the animal experiments. Similarly, the ratio of the area under the curve for MDIF to IDIF in the human experiments was 1.05 ± 0.11, resulting in agreement in grey matter CMRGlu: 24.2 ± 3.5 and 24.4 ± 3.8 mL/100 g/min for MDIF and IDIF, respectively. The MDIF method was superior in characterizing the first pass of [ 18 F]FDG. Groupwise macro- and microparameter voxelwise images obtained with the MDIF showed the expected patterns. Conclusions: A model-driven SIME method was proposed to derive high SNR input functions. Simulations indicate that [ 18 F]FDG rate constants obtained with the MDIF had higher precision than corresponding estimates from the original IDIF SIME approach, which is attributed to the MDIF method requiring fewer fitting parameters to characterize the input function.