Assessment of the Mosaic animal PET system response using list-mode data for validation of GATE Monte Carlo modelling

Monte Carlo modelling is widely used in nuclear medicine for system optimization and for image processing in order to obtain system data that cannot be measured experimentally. Any simulation model should be validated, especially in terms of spatial resolution, count rates, scatter fraction and sensitivity in the case of positron emission tomography (PET) systems. The aim of this study was to compare the performance of a Mosaic animal PET system model to the real system by using raw data in a list-mode format, in order to preserve all the precision and accuracy of the acquired data. GATE package was used to design a realistic model of the Mosaic system. Detectors, shielding, cap, air medium, phantoms, sources and electronic processing were simulated in GATE. Sensitivity obtained from experimental and simulated data were compared using the list-mode data for different energy windows. Measurements of count rates were performed for uniform cylindrical phantoms for both mouse and rat. Output data were recorded with the same spatial sampling as the experimental data. Point spread functions at different locations in the field of view were analysed in raw and reconstructed formats. Simulated and measured sensitivity differed by less than 5%. Simulated and measured single and prompt count rates agreed within 6% for activities up to 100MBq for the two phantoms. A semi-empirical approach was used to simulate energy efficiency and losses of resolution after crystals interactions (detector blurring). Spatial resolution, assessed on simulated sinograms and on reconstructed data, agreed with real data. Comprehensive evaluation of the Monte Carlo modelling of a microPET system was performed using list-mode data and showed that the GATE model was appropriate to accurately reproduce the response of the system.