[Gate-users] Simulation killed

[florian.bretin at ulg.ac.be]

Dear Gate users & experts, 

I am simulating a microCT acquisition of 220 projections over 193° of rotation. A PMMA phantom (Analyze7.5 image) was implemented as “ImageNestedParametrisedVolume” into the center of rotation and is rotated instead of the gantry. After the simulation I collect the dose profiles for analysis inside the phantom from the dose map. 
The simulation is run with 6e+07 photons / sec and 16ms exposure per projection (total simulation time of 3.52). Im using Gate v6.1 on Ubuntu 11.10, 22Gb of ram and Intel Xenon 6 core X5670.

Here is a link to the simulation files:
https://www.dropbox.com/s/2mt6kchejnho0qi/microCT.rar

Now I have two problems:

1.	The simulation runs fine with the X-ray source being in the center position of the phantom. However, when I translate the phantom 40mm into positive X-direction, the simulation is killed after 177 projections. The error message I get is simply “Simulation killed”, no error code. During projection 177 the memory piles up completely (including Swap) until the simulation is killed. I have googled the problem including the mailing list but did not find a solution. Did anyone ever face a similar problem? Any ideas? 

2.	The second problem is, that I have dose deposition in air outside of the phantom, which I cannot explain. I am simulating less photons than the actual tube emits and I scale the resulting dose maps to experimental results in order to reach the number of photons the tube emits. My wild guesses are: 1) amplification of noise due to the scaling, 2) scatter from the phantom due to the sharp edge.


Here is a link to the dose profiles:
https://www.dropbox.com/s/z3q35cekb0ekr59/GateExampleDoseProfile.tif
https://www.dropbox.com/s/ly16idlbmqndkkq/PhantomHUvsDose.tif

Any help / ideas would be much appreciated.

Thanks in advance & kind regards
Florian Bretin 
Cyclotron Research Centre, University of Liege, Belgium