Example code of running GPUMCI on realistic skull phantoms (nifit-format) on GPU-cluster. Project part of the KTH-SSF-LCR-project.
Main run file. Imports x-ray spectrum, phantom, sets up geometry, potential gain, phantom material definition, etc. and simulates using GPUMCI (in-house MC-based simualation package). Embarrassingly parallel simulations setup for GPU-cluster.
Bash script to run HelicalCT_Head_Geriatrics.py in parallell on GPU-cluster
Standard spectrum (W-anode, 120 kVp, 0.05 rel. voltage ripple, air kerma: 66*10^-6 Gy, mean energy: 64.5 keV, window shielding: 30 mm Be, 7 mm Al, 0.1 mm Cu) used for head simulations. File given in two column form, 1st column: keV, 2nd column: intensity. Values are normalized when imported in HelicalCT_Head_Geriatrics.py.
Definng path to skull phantoms on KTH imagingnas (files not include on github because of 1) proprietary nature, and 2) file size).
Script to append output files from HelicalCT_Head_Geriatrics.py. Script sums upp all simulations, but conserves dicsrete turns, i.e. raw output files on the form
mySim_Sim_num_0_Turn_num_0.py
mySim_Sim_num_1_Turn_num_0.py
mySim_Sim_num_2_Turn_num_0.py
.
.
.
mySim_Sim_num_N_Turn_num_0.py
mySim_Sim_num_0_Turn_num_1.py
mySim_Sim_num_1_Turn_num_1.py
mySim_Sim_num_2_Turn_num_1.py
.
.
.
mySim_Sim_num_N_Turn_num_1.py
is converted to
mySim_all_Sim_num_Turn_num_0.py
mySim_all_Sim_num_Turn_num_1.py
Substances used in GPUMCI, with white_matter, gray_matter, and bone stochiometric composition defined and ready for attenuation table generation in GPUCMI.
To reconstruct generated data, feel free to use the code found in ad-skull-reconstruction.
To run the above, you have to have odl (with dependencies) installed, along with xraylib (for generating attenuation tables) and GPUMCI (KTH in-house software. Contact @davlars or @adler-j for more information).