[Castor--users] TOF image reconstruction using GATE simulation data

Dear CASToR users,

I try to reconstruct the TOF image by converting the root result obtained by the GATE simulation into a castor file.
I have added the TOF_reso command and converted the data. Do I need any other commands besides this command?

The command I wrote is below.
castor-GATERootToCastor -i my_data.root -m my_data.mac -o my_TOF_image -s my_data -TOF_reso 220. -src
In the above command, TOF image and nonTOF image were reconstructed using TOF_reso command. However, image results were nonTOF image better than TOF image.

The commands used to create the two images are shown below.

castor-recon -vb 2 -df 181221_nonTOF_t15_60s_jaszczak_src_df.Cdh -fout 181221_nonTOF_t15_60_jaszczak_src -it 5:5 -dim 400,400,400 -opti MLEM -proj joseph -conv gaussian,3.,3.,3.::psf

How can we solve it?

Any help I would really appreciate it.

Best regards,

Yeongkyeong KIM

Hi Yeongkyeong,

There is no additional commands to use besides -TOF_reso, which will provide the assumed time resolution of the system. The TOF measurements dt for each event is computed from the time1 and time2 ROOT variables, and the time measurement range (written in the datafile header) is recovered during the conversion process. You could check if the dt measurements in your datafile seems consistent using the castor-datafileExplorer executable with the following options to explore the datafile event by event : -df /path/to/datafile/header -i -e
Depending on the ToF resolution, the convergence with ToF correction could be much faster than with non ToF. At some point, increasing the number of iterations would just increase the noise level, so it could be worth to check the image you get just after a few iterations (e.g 3, 4 iterations, without subsets).

Hope this helps !

Kind regards,

김연경 <dusrud026@gmail.com> a écrit :

Dear, CASTor user

First, thank you for your reply. Merlin

I reconstructed the TOF image using the benchmark example. There were two example files.
One is list-mode data, and the other is histogram-mode data.
However, when I looked at the header file of both data, they were both TOF data.
The differences found in the two header files were the number of events different. The number of counts of list-mode data was about 1.5 times larger.
Why does this make a difference?

I compared non-TOF data obtained by Gate simulation and converted images into list-mode and histogram-mode, respectively. In conclusion, images using histogram-mode data were better.
Could you explain the exact reason?

Any help I would really appreciate it.

Best regards,

2019년 1월 4일 (금) 오전 1:05, Thibaut Merlin <Thibaut.Merlin@univ-brest.fr>님이 작성:

Hello Yeongkyeong,

Both PET benchmarks were indeed contain TOF data. For histogram data, the number of events corresponds to the number of histogram bins, and not the number of counts as for list-mode data.

Could you indicate what is different between the images reconstructed from the histogram and list-mode datafiles generated from your GATE simulation data ? The resulting images should be identical, but there could be some small differences due to the way the sensitivity image is computed in both case.