Development of 3D dynamic acquisition in a neuro-PET scanner

The acquisition of 3D septa-less data in PET is gaining increasing acceptance but is still hampered by large data volumes and relatively long data transfer times. These problems are compounded by the desire for multiple time frames of data so that the increased efficiency of 3D can be fully utilized to define dynamic physiological processes. The tracer kinetics of ¹¹C-diprenorphine and ¹¹C-flumazenil (opiate and benzodiazepine receptor ligands respectively) have been studied using up to 20 time frames of 3D data. If the maximum number of projections (192) plus dual energy window data (for scatter correction) are acquired (32 MBytes per frame), the minimum frame duration in our system is 3 min. This limit is imposed by the available sorter buffer memory (128 MBytes) and the disk transfer time (0.2 to 0.4 MBytes per sec). Although adequate for the measurement of volume of distribution (V_D), for instance, this minimum time frame would not be so for many other dynamic studies. The effects of reducing the volume of data, by (i) averaging of projections and (ii) spatial averaging of the lower energy window, on resolution, scatter correction and V_D have been investigated. Preliminary findings indicate that there is considerable scope for data reduction without significantly affecting the final result.