Extraction of Input Function from Rat [18F]FDG PET Images

Abstract

Small animal positron emission tomography (PET) with 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) facilitates the visualization and quantification of glucose uptake in rats and mice. The quantification of glucose uptake requires an input function, which is generally obtained by measuring radioactivity in arterial plasma withdrawn during PET imaging; however, this approach is not always feasible because abundant blood sampling may affect the physiological process being measured. The purpose of the present study was to develop a new model-based technique (K-Model) and compare it to the previous F-Model. The study material consisted of two separate groups of rats having different physiological conditions. Each group was scanned by different PET cameras, i.e., HRRT and Inveon-PET/CT, and blood samples were drawn during imaging. Two kinds of model functions, i.e., F-Model and K-Model, were used for estimating input functions by an optimization procedure, applying restrictions on boundary conditions. To validate the method, glucose influx rate, Ki, was computed from the estimated and measured input functions for comparison. The input functions were well reproduced when single-point blood count data were used for both models. The difference in Ki values between the model-based and blood sampling methods was 1.1±15.1% by K-Model which showed the most feasible in the study. The regression analysis showed a tight correlation between the image-based and blood sampling methods, and the slope was close to unity and the intercept close to zero. It is possible to estimate the input function from rat [18F]FDG PET images, thus facilitating the assessment of glucose metabolism without affecting the physiological conditions of the animal as a result of abundant blood sampling.