Validation of an analytic method of calculating cerebral glucose metabolism using PET

Quantitative modeling of cerebral metabolic rate for glucose (CMRglc) using PET with the FDG method requires calculation of the integral of the time course of radioactivity in arterial plasma. Numeric integration has typically been used but requires 30 or more blood samples taken between 15 s and 100 min after injection of the radiotracer. Our laboratory has developed an alternative integration method that fits the values of the plasma samples to an analytically integrable function using only 4-6 samples taken between 40 and 110 min after radiotracer injection. Methods: The plasma integrals were calculated by both the analytic and the numeric methods with data from FDG PET studies that were not used in the development of the analytic method. In 39 PET studies from 22 healthy volunteers, 30 plasma samples were taken over 110 min. Results: The plasma integrals determined by the analytic and numeric methods yielded a within-subject correlation coefficient of >0.95 and differences of < 10%. Conclusion: Because the analytic method requires less blood sampling and does not require sampling immediately after radiotracer injection, the experimental procedure is simplified without loss of accuracy in CMRglc computation, and the effect of missing or incorrect samples is reduced.