Experimental validation of a PET Monte-Carlo simulator

As with most PET simulators, the widely used SimSET (a Simulation System for Emission Tomography) package still needs to be thoroughly validated. This paper presents the continuation of a comparison between simulations and experimental validation data obtained on two cylindrical SIEMENS tomographs: the ECAT HR+ and the ECAT ACCEL. The 2D and 3D sensitivities were simulated. The standard 2D collimator module did not give good results, so we described the 2D collimator inside the phantom. As the standard version of SIMSET does not describe block detectors, the sensitivity is corrected by the square of the ratio between the volume of the detector blocks and the volume of the full detector ring. The agreement is good for 2D and 3D as well for the HR+ tomograph, a little less satisfactory for the ACCEL tomograph. The standard SimSET computes the centroid of the photon interaction in the detector ring. As there is no Depth Of Interaction measurement on these machines, the spatial resolution was simulated with the hit in the detector projected onto the input side of the detector. Thus, with this simple correction, the spatial resolution is in correct agreement with experimental data, the discrepancy increases of course if quite far from the axis. Most simulations were done with and without variance reduction techniques. The agreement between them is good. Complementary comparisons are presented, such as energy spectra of a water cylinder phantom outside the Field Of View for the ACCEL machine.