[Gate-users] Divergence between GATE simulation and the reality

[Maxime Toussaint]

Honorable Zhengzhi,

Good luck about your task. Keep in mind that with Gate we often define a digitizer for the whole system while a real scanner could have module-dependent characteristics which might explain some of the difference.

About the  Lu167, I believe (Please correct me if I am wrong anyone!) that only explicitly defined "source" can generates primaries(I think that is the term?) by themselves. Thus, you could fill your scanner with Uranium material and it would not generates anything by itself. I am not a physicist, so I do not know what Lu167 does and how. However, it should be doable with the appropriate physic list.

Have a nice day,
Maxime Toussaint

________________________________
De : Albert Grace Lieu <albertnew2018 at gmail.com>
Envoyé : 29 mai 2020 23:58
À : Maxime Toussaint <Maxime.Toussaint at USherbrooke.ca>
Cc : gate-users <gate-users at lists.opengatecollaboration.org>; Peter Olcott <polcott at reflexion.com>
Objet : Re: [Gate-users] Divergence between GATE simulation and the reality

Honorable Maxime,

Thank you very much for your reply and your knowledge. You always give the best suggestion and answers. I definitely have spent much time on digitizer setting, such as singles readout level, deadtime (and deadtime volume), delay, coincidence window and delay window width, function modality (non/paralysable ). I experimented with these parameters for many times, and also validated it against our real data. From the comparison, these settings seem to be alright, but the GATE model gives me more counts  (~25%) than the real system.

One thing worth noticing is that the NEMA sensitivity experiment uses a very low dose (500 uCi). Under such a low count rate, it sounds easy to figure out why GATE has more counts because detector is not saturated and deadtime bearly kick off,  but I have been stuck for about half a month already on this point.

Another point is that does GATE simulate Lu167 decay/deexcitations in its LSO and LYSO material when the material is called in GATE? If not,  how can I add it to my simulation? Sorry about asking this question before doing my own research on this topic.

I will update this topic if I find something interesting regarding this discrepancy between GATE simulation and the real system.
Thank you very much for your time and excellent expertise.

Cheers.
Have a mellow weekend.

Zhengzhi

On Fri, May 29, 2020 at 7:52 PM Maxime Toussaint <Maxime.Toussaint at usherbrooke.ca<mailto:Maxime.Toussaint at usherbrooke.ca>> wrote:
Greetings,

That is a tricky question. Sadly, I do not know if such a convenient procedure is available within Gate.  While I do not have much experience in modeling an accurate acquisition system, I can offer a couple of ideas which might be useful or not.

In my humble belief, the trickiest part of your task is to define the digitizer [1] and I think it might also be where the discrepancy appear. If you want a realistic model, you should have a pretty complex digitizer that exploit most of the available module. For example, having underestimated the Dead Time could explains what you see. It could also be due to a mismatch in the depth designation of the digitizer module: setting the DeadTime on crystal level allow more events then allowing it on module level. Thus, the only advice I can give you is to make sure that your digitizer configuration correspond to the scanner by validating each module if possible. If any of the digitizer module do not have what you need, maybe some more advanced Gate Users could provide a solution!

Have a nice day and good luck,
Maxime Toussaint

[1] https://opengate.readthedocs.io/en/latest/digitizer_and_detector_modeling.html?highlight=pileup#digitizer-modules<https://can01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fopengate.readthedocs.io%2Fen%2Flatest%2Fdigitizer_and_detector_modeling.html%3Fhighlight%3Dpileup%23digitizer-modules&data=02%7C01%7CMaxime.Toussaint%40usherbrooke.ca%7Cf23932b020fe4fe8920d08d8044dd31c%7C3a5a8744593545f99423b32c3a5de082%7C0%7C0%7C637264079583608384&sdata=uwY9McP%2F3mEIQDnhYI7%2BJAproGNT9N6OR9aCXXaQLGA%3D&reserved=0>


________________________________
De : Gate-users <gate-users-bounces at lists.opengatecollaboration.org<mailto:gate-users-bounces at lists.opengatecollaboration.org>> de la part de Albert Grace Lieu <albertnew2018 at gmail.com<mailto:albertnew2018 at gmail.com>>
Envoyé : 29 mai 2020 13:32
À : gate-users <gate-users at lists.opengatecollaboration.org<mailto:gate-users at lists.opengatecollaboration.org>>
Cc : Peter Olcott <polcott at reflexion.com<mailto:polcott at reflexion.com>>
Objet : [Gate-users] Divergence between GATE simulation and the reality

Dear GATE users,

I have modeled a real PET system in GATE as close to the real system as I can. However, during the noise equivalent count collaboration, GATE is always much better (about 50%) than the real system. After lots of effort, we looked at the results of the NEMA sensitivity experiment (very low dose, ~500 uci) and found that GATE collected about 25% more counts than the real system.

I was wondering if there is some calibration work done between GATE simulation and reality. Could someone give any help on how to further debug or make clear what causes this discrepancy between simulation and reality?

Thanks a lot.
Zhengzhi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opengatecollaboration.org/pipermail/gate-users/attachments/20200531/cddf6498/attachment-0001.html>