Calibrated fractional uptake rate to estimate cerebral glucose consumption: a simplified approach less sensitive to inter-frame motion than patlak analysis

Long acquisition times and resulting inter-frame patient motion may compromise accurate measurement of cerebral metabolic rate of glucose (CMRGlu) with [18F]-Fluorodeoxyglucose (FDG). To overcome this limitation, the present study compared the CMRGlu estimations provided by the compartmental analysis with the ratio between late brain FDG uptake and integral of the input function, i.e., the Fractional Uptake Rate (FUR). The study included 12 patients without known cognitive impairment undergoing FDG-PET for suspected lung cancer. Image Derived Input Function (IDIF) was obtained from the descending aorta using a dynamic acquisition centered on the heart. Thereafter, the cohort was divided into two groups submitted to a dynamic acquisition either centered on the head (ACH, n = 6) or whole-body in continuous bed motion (CBM, n = 6). Images were normalized to a PET template and segmented into 83 brain regions for quantitative analysis; K i values were multiplied by serum glucose and normalized for the lumped constant to compute CMRGlu. FUR systematically overestimated IDIF-based Patlak values by approximately 32%, indicating a substantial quantitative bias; however, this overestimation decreased over time after tracer injection. Despite this bias, whole-brain FUR- and IDIF-based Patlak-derived CMRGlu showed a strong linear relationship (R 2 > 0.80), which was consistently preserved across brain regions. A pooled calibration markedly reduced the relative error between approaches, and its performance was confirmed by leave-one-subject-out cross-validation. Unlike Patlak analysis, FUR-derived CMRGlu were minimally affected by inter-frame motion-correction. Although intrinsically time-dependent, FUR shows a strong linear relationship with IDIF-based Patlak estimates and, after appropriate calibration, improved quantitative agreement. Its lower sensitivity to inter-frame patient motion may enhance the clinical feasibility of quantitative brain FDG-PET, particularly when full dynamic acquisition is difficult to perform. Its main practical advantage lies in simplifying routine patient acquisition, provided that an initial calibration and validation procedure has been performed. NCT02475382; registration date: 18 June 2015.