A semipopulation input function for quantifying static and dynamic 18F-fluoride PET scans.

Purpose: We describe a semipopulation input function for evaluating bone plasma clearance from static and dynamic (18)F-fluoride PET scans.

Methods: The semipopulation input function was derived by fitting an exponential curve to venous plasma measurements obtained 30-60 min after injection and adding a population residual curve representing the bolus peak scaled for injected activity and adjusted for time of peak counts. The residual curve was found from nine postmenopausal women who had continuous arterial blood samples and venous samples taken every 10 min. The precision errors of plasma clearance measurements derived from the semipopulation input function using Patlak analysis and the Hawkins compartmental model were compared with the precision errors for four image-derived input functions using data from 20 women who had undergone repeated dynamic PET scans.

Results: Venous and arterial concentrations were equal by 30 min after injection. The exponential fitted to the 30-60-min venous data accounted for 76% of the total 0-60 min area under the curve, and the SD of the area under the residual curve was 2.6% of the total 0-60 min area under the curve. For Patlak analysis, the precision error (% coefficient of variation) was 13.0% using the semipopulation input function compared with 14.9-21.7% using the four image-derived input functions. For the Hawkins model the equivalent figures were 14.5 and 20.1-30.9%, respectively.

Conclusion: Accurate and precise measurements of bone plasma clearance were obtained when (18)F-fluoride PET scans were analysed using an input function obtained by adding a population residual curve to the exponential obtained from venous blood samples taken 30-60 min after injection.