Quantitation in 3D brain imaging using gamma camera PET

This work presents methods developed to perform quantitative FDG brain imaging with a dual-headed gamma camera (Picker 2000XP) adapted to perform PET acquisitions. Noise equivalent counts analysis, including activity outside the field of view, indicates that 3D acquisition produces superior data and facilitates a lower patient dose than acquisition with septa. To preserve the resolution in this 3D data set, a 3D image reconstruction algorithm is required. Consequently, a novel 3D backproject then filter algorithm has been developed which utilises all of the measured data. This algorithm offers superior spatial resolution to that provided by the system software, which is based on single slice rebinning. The most important data correction is shown to be attenuation correction. Due to the current lack of transmission scanning, attenuation correction is performed using a non-uniform attenuation map derived from the emission data. Monte Carlo modelling is being used to optimise a lower energy window scatter correction method. As the system is not provided with a delayed coincidence window, random correction is achieved using a fast Monte Carlo method. The gamma camera PET images were validated through comparison against a GE Advance scanner for the same patients. Region of interest analysis revealed a mean ROI discrepancy of 3.8% using the new software, compared to 8.0% using the system software