Patient-specific noise-equivalent-counts from repeated, dose varying [O-15]H₂O PET scans

Scanning protocols in dynamic PET imaging can be optimised to improve the accuracy and precision of measurements of physiological parameters. For [O-15]H₂O scans, one of the major factors that can be controlled is the injected dose, which affects the statistical quality of the PET data as well as its accuracy. We build upon a previous dose optimisation methodology that considers patient-specific noise-equivalent-count-rates (NECRs), extending its use to dynamic PET scanning. The accuracy of this technique is examined within 4 individual patients. Each of these patients were given a series of between 4 and 6 injections of [O-15]H₂O at doses that were intentionally varied between 140 and 740 MBq. Patient-specific NECR-dose curves were extrapolated from each injection. These data are used to qualify the performance of NECR-dose curve predictions in a dynamic PET setting, as found on a modern LSO PET/CT scanner. The optimal dose was defined as the dose required to give 95% maximisation of the integral NEC within a specified time interval post administration. We find good agreement between the extrapolated NECR curves within an individual. On average, the variability between injections produced a standard deviation of 3 % in the optimal dose calculated for the specific patient. The optimal doses ranged from 264 to 640 MBq between subjects, for maximising the NECR during the times most critical to the measurement of blood flow. We also find that although administration of the tracer at the back of the wrist in a pelvic scan produces a higher global NECR, the signal-to-noise ratio in the image is reduced compared to a similar injected dose at the elbow. We conclude that patient-specific NECR methods can provide useful guidance in dose optimisation.