I have a question regarding the reconstructed images. I tried to reconstruct GATE simulation data from my own scanner setup. The images looks well. However, there is a problem with voxel values and units.I performed attenuation correction (attenuation map in cm^-1) and sensitivity correction (long list-mode acquisition with high number of true coincidences). I prepared new normalization map which merged attenuation and sensitivity (-opti SENS recon option) and put that as an -img input for the main reconstruction. What I found is that in case of uniform distributed phantom the voxel values are quite low (i.e. 2*10^-8), however the background values dropping down every iterations (i.e. from 10^-2 - 2nd iteration up to 10^-34 15 iteration). I am wondering what kind of calibration factor should I apply to get proper units i.e. number of counts ? How should I eventually calculate that factor? Is it somehow dependent from the iteration number?
I need it to perform NRMSD (Normalised Root Mean Square Deviation) analysis to find the optimal number of iterations.
All the best,