Towards Quantitative Image Reconstruction Using Monte-Carlo Simulations in Multi-Wire Proportional Chamber-Based Small Animal PET

The quadHIDAC is a dedicated high resolution small animal PET scanner based on multi-wire proportional chamber detectors. It provides very high spatial resolution, good sensitivity, and a large field-of-view allowing the examination of more than one animal in a single scan or a whole body scan of larger species in a single bed position. Excellent image quality has been obtained through list-mode-based EM reconstruction using resolution recovery. However, quantitative results on local uptake are still missing due to both (1) lack of energy discrimination and (2) inaccurate characterization of random events recorded by the scanner. In this work we describe how quantification can be achieved through the use of an appropriate Monte Carlo simulation of the quadHIDAC. The complex structure of the quadHIDAC has been completely implemented in a Geant4 simulation. Point sources, line sources, cylindrical phantoms, and voxelized image data have been successfully simulated and compared to real measurements. The magnitude and spatial distribution of scattered and random events could be quantitatively characterized for the different phantom geometries. The simulation provides access to information on single hit distributions resulting in appropriate randoms corrections in image reconstruction. Both randoms and scatter correction is demonstrated to improve quantification. Their accuracy and performance was tested on simulated and measured data. Using simulation-based correction techniques for quantitative image reconstruction, multi-wire proportional chamber-based PET technology may be attractive for future high resolution PET scanners.