[Castor-users] Bug toolkits/castor-GATERootToCastor.cc

[tmerlin]

Hi,


Thank you both for your feedback!


Indeed there are several issues in the converter occurring when one 
would like to enable ToF and integrate some custom scatter/random 
corrections.


I hadn't checked /castor-GATERootToCastor/ in v3.0 with ToF option 
enabled, and as Jan experienced, it crashes because of incorrect 
initialization of the ToF measurement range for the projector classes.


Additionally, as Maxime suggested, it actually makes no sense to use ToF 
in the projector to compute ACF.


-> Quick workaround is to perform the modification Jan suggested, and 
also add the following line in red after the ProjectorManager 
initialization:


"

     ...

     // Check parameters
     if (p_ProjectorManager->CheckParameters())
     {
       Cerr("***** castor-GATERootToCastor :: A problem occurred while 
checking projector manager's parameters !" << endl);
       Exit(EXIT_FAILURE);
     }

     // Initialize projector manager
     if (p_ProjectorManager->Initialize())
     {
       Cerr("***** castor-GATERootToCastor :: A problem occurred while 
initializing projector manager !" << endl);
       Exit(EXIT_FAILURE);
     }
p_ProjectorManager->SetSensitivityModeOn();

   ...

"


This will force the projector to use the non-ToF projector functions anyway.


-> Regarding the correct initialization of the ToF measurement range, I 
agree that it should be initialized with the system coincidence window, 
rather than derived from the data as Marina and Maxime suggested. This 
will be fixed in next update (with an additional option to manually set 
it). A temporary workaround for this is to set it directly in the castor 
datafile header once it has been generated.


-> Regarding the reconstruction of v.2 CASToR datasets with the v.3 
code, there should not be any issue. The error message comes from 
additional check in the v3 version.

However, the estimation of ACF during conversion was already performed 
by the projectors with ToF enabled, so the ACF values might be wrong!



Thank you all for your insights!


Best wishes,
Thibaut






On 23/01/2020 16:56, Maxime Toussaint wrote:
> Greetings,
>
> I was curious about your discovery and started to check it out. I am 
> not totally sure but I think there is something else in this bug.
>
> The tool castor-GATERootToCastor only require a projector object when 
> evaluating attenuation coefficients. Since these coefficients are 
> independent of ToF, the projector should not need ToF information. 
> Worse, if the projector use the ToF information, the attenuation 
> coefficients will be evaluated from a portion of the LOR which does 
> not fit the reality. However, my previous statement might not be true 
> for on-the-fly system matrix computing. I will leave it to more 
> knowledgeable peers to confirm or infirm that part.
>
> In short, my understanding is that ToF should be disabled for the 
> tools castor-GATERootToCastor when evaluating attenuation coefficients 
> (Using the SetIgnoreToFFlag method, most likely) which should remove 
> the error you observed.
>
> On that note, I also like the idea that the variable 
> m_ToFMeasurementRangeInPs can be user/scanner defined. While the 
> current approach defined in castor-GATERootToCastor, which is data 
> dependent(1), is efficient, it is true that we sometime want to fix 
> that value for comparison or whatnot.
>
> Have a nice day,
> Maxime
>
> (1) ((iDataFilePET*)Out_data_file)->SetToFMeasurementRangeInPs(dt_max 
> - dt_min);
>
>
>
> ------------------------------------------------------------------------
> *De :* Castor-users <castor-users-bounces at lists.castor-project.org> de 
> la part de Grahe, Jan <jan.grahe at pmi.rwth-aachen.de>
> *Envoyé :* 23 janvier 2020 07:34
> *À :* castor-users at lists.castor-project.org 
> <castor-users at lists.castor-project.org>
> *Objet :* [Castor-users] Bug toolkits/castor-GATERootToCastor.cc
>
> Dear CASToR developers,
>
> I tried converting a PET list-mode generated by GATE into the CASToR 
> format using the castor-GATERootToCastor toolkit and obtained an error:
>
> “….
>
> sScannerManager::Initialize() -> From scanner PET_PHILIPS_VEREOS
>
> iScannerPET::Initialize() -> Initialize remaining stuff for scanner to 
> be ready
>
> oProjectorManager::Initialize() -> Initialize projectors and 
> projection lines
>
>   --> Use ToF projector for listmode data
>
> ***** vProjector::CheckParameters() -> Inconsistent ToF related 
> parameters !
>
> ***** oProjectorManager::ParseOptionsAndInitializeProjectors() -> A 
> problem occurred while checking forward projector parameters !
>
> ***** oProjectorManager::Initialize() -> A problem occurred while 
> parsing projector options and initializing it !”
>
> Turns out, that the variable /m_ToFMeasurementRangeInMm/ has a value 
> of “-0.1498962” in the /oProjectorManager/, which caused the error in 
> the /vProjector::CheckParameters()/. This variable is obtained from 
> the /vDataFile* Out_data_file/ using the method 
> /GetToFMeasurementRangeInPs()./ The datafile is initialized in the 
> castor-GATERootToCastor.cc file and handed to the /p_ProjectManager/ 
> using the /SetDataFile()/ method. However, the 
> m_/ToFMeasurementRangeInPs/ variable is set to -1 in the constructor 
> and not set in the /castor-GATERootToCastor/ toolkit before the 
> /oProjectorManager/ requires this variable to be set when calling it’s 
> /Initialize() /function.
>
> Therefore, to use the toolkit with ToF right now I added the red line 
> in the toolkit (the block is approx. at line ~970 of the document):
>
> “….
>
> // Set ToF
>
>       if(ToF_reso>=0)
>
>       {
>
>         ((iDataFilePET*)Out_data_file)->SetToFInfoFlag();
>
> ((iDataFilePET*)Out_data_file)->SetToFResolutionInPs(ToF_reso);
>
> ((iDataFilePET*)Out_data_file)->SetToFMeasurementRangeInPs(myScannersToFrange);
>
>       }
>
> ”
>
> This seems to solve the problem, but only with the _general 
> coincidence time window of the scanner_ instead of the longest 
> coincidence time of all events in your datafile. Not sure if that is 
> correct or a factor of e.g. 2 has to be applied. If there is no 
> problem with that, you could fix this bug generally by getting the 
> coincidence time window of the scanner and then later update it to the 
> real variable when all the events have been processed.
>
> As this only happens in version 3.0, can I still reconstruct images 
> with CASToR 3.0 using datafiles generated with the CASToR 2.0 framework?
>
> Thanks and I hope this helps!
>
> Jan
>
>
> _______________________________________________
> Castor-users mailing list
> Castor-users at lists.castor-project.org
> http://lists.castor-project.org/listinfo/castor-users

-- 
Thibaut MERLIN -- PhD

Docteur en Imagerie Médicale au Laboratoire de Traitement de l'Information Medicale (LaTIM - INSERM UMR 1101)
Institut Brestois de recherche en Bio-Santé (IBRBS)
12 Avenue Foch, 29200 Brest, FRANCE
Tel: 06.75.12.24.90

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.castor-project.org/pipermail/castor-users/attachments/20200124/b447897a/attachment.html>