loading data using GUI

[milad najafzadeh]

Dear experts,

I changed my pipeline to use the cmp3 and have a question regrading DWI data types when loading the bids app ( DTI or multi-shell).  My DWI data were acquired with a b0 volume (one and only one) and 64 volumes with b=1000.  So which of the options (DTI or multi-shell) provided by bids GUI must be chosen? Given that these data principally are considered as a single shell. 

Best regards,

-Milad

[Sebastien Tourbier]

Dear Milad,

In your case with one shell, I would select DTI. You would still be able to select mrtrix3 CSD for reconstruction of the diffusion signal.

The only difference with multi-shell is that multi-shell allows you to use in addition MAP-MRI to compute extra maps. But this requires to have data acquired with multi-shell scheme.

Hope this helps.

Let me know if have further questions.

Best,

Seb

[milad najafzadeh]

Dear Sebastien

 

Many thanks for your response. That was very helpful.  I have also a couple questions regarding the mrtrix CSD model and ACT. Since my data are single shell, a response function can be obtained for preforming (called kernel) CSD model using the dwi2reponse function. In this case, which of dwi2reponse algorithms are used (https://mrtrix.readthedocs.io/en/latest/reference/commands/dwi2response.html)?  

For performing ACT, the required step is correction of EPI susceptibility distortions. The data have no reversed phase encoding direction and topup can’t be performed. I use this method (https://www.frontiersin.org/articles/10.3389/fninf.2017.00017/full) for solving this issue and then perform ACT. In the  cmp3, such transformation (non-linear between T1w and b0 volume) can be perform using ANTs. Am I right?

 

Sorry for long questions but I need to clarify these issues for myself

 

Thanks in advance to your helps,

 

Best regards,

 

-Milad

[milad najafzadeh]

Dear Sebastien

 I am looking forward to your answer, please help me.

Best regards,

 

-Milad

[Sebastien Tourbier]

Dear Milad,

Sorry, just discovering that there were some new messages in the group.

This group has been created at the time of the first CMP but I do not really like to software discussion and code managed in two different places, that might make the finding of existing information more difficult.

The good news is that GitHub also provide a Discussion tab where questions can be asked and answered: https://github.com/connectomicslab/connectomemapper3/discussions

I really encourage you to use it in the future and forget about this group. Your questions / bugs are going to be discovered and addressed much faster ;)

Please find my answers to your 2 previous questions:

- The response function that is used for CSD uses the tournier algorithm, so it is appropriate for your data. In case, this corresponds to this part in the CMP3 code:

  https://github.com/connectomicslab/connectomemapper3/blob/9126faa155d3cada2fea3630ba3e589dc1b72ba7/cmp/stages/diffusion/reconstruction.py#L649-L661

- Yes, when performing ACT, it is really important to have all non-linear distorsions corrected. This is indeed the purpose of a non-linear co-registration of DWI b0 and T1w, that is contrained along the EPI distorsion direction. This applies to data with no fieldmaps or to data that is not acquired in the two phase encoding directions.

Hope this helps!

Best,

Seb

[milad najafzadeh]

Dear Sebastien

Many thanks for your detailed answers

With kind regards,

-Milad