Newbie, Registering question and doubts

Dear experts,

I am just starting using MRTrix and I have some questions, I really would appreciate any help.
My image is a MultShell DWI (9b0, 32 b1000 and 32 b2000), with additional inversed phase encoding direction T2 image and a T1-3D.

1 - I would like to register my T1 image with my post-processed DWI (I understand that this is not mandatory). I noticed that the “dwifslpreproc” function corrects the diffusion volumes for susceptibility artifacts, but not the b0 images, am I right? If so, how to get corrected b0 volumes? ( I am guessing this would be the best approach for FLIRT and FNIRT).

2 - Besides the “dwifslpreproc”, do I need to run the “dwipreproc” also?

In the case I am lucky, maybe you may comment on my pipeline:

dwidenoise
dwifslpreproc (using the inversed phase T2)
dwibiascorrect
masking
dwi2response dhollander
dwi2fod msmt_csd

Thank you very much!

Hi,

Your multishell sequence was acquired all in a single acquisition? Or do you have the different shells saved as different images? I will assume for the responses that everything is together. When you say inversed phase encoding direction T2 image, are you refering to an additional B0 with the opposite phase encoding direction that you main diffusion acquisition, am I right?

1 - I would like to register my T1 image with my post-processed DWI (I understand that this is not mandatory). I noticed that the “dwifslpreproc” function corrects the diffusion volumes for susceptibility artifacts, but not the b0 images, am I right? If so, how to get corrected b0 volumes? ( I am guessing this would be the best approach for FLIRT and FNIRT).

dwifslpreproc will correct the B0 as well, you don’t need to worry about this. It will correct your full multishell dwi.

2 - Besides the “dwifslpreproc”, do I need to run the “dwipreproc” also?

No, dwipreproc is an older version of dwifslpreproc.

In the case I am lucky, maybe you may comment on my pipeline:

dwidenoise
dwifslpreproc (using the inversed phase T2)
dwibiascorrect
masking
dwi2response dhollander
dwi2fod msmt_csd

Your pipeline looks fine, you could add mrdegibbs to remove the gibbs ringing artifact (after dwidenoise). And as a final step, I will run mtnormalise. Depending on your objective, you need to use the same response function for all your subjects. I hope this helps.

Best regards,

Manuel

1 Like

Dear Manuel,

Thank you very much for your precise reply. You were right in all assumptions: that is a Single acquisition Multi-Shell, plus a separated single b0 (what I meant by pure T2) with phase encoding in PA direction.
I will try to figure out why my b0 images still look (very) affected by susceptibility after dwifslpreproc.
Thank you also for the suggestions in the pipeline, I will include them all.
Have a nice day!

Hi Brunno,

Your external B0s won’t be corrected, but the ones included in the multi-shell acquisition should be corrected. If this is not the case, something is going wrong.

Best regards,

Manuel

1 Like

Hi Manuel, thank you.

My post processed b0 images were weird because my b0 image with inverse phase encoding direction is actually a b0 + 6 b1000 image, so, the dwifslpreproc probably averaged b0 and 1000 to correct SuscArtif. and everything went wrong. I extracted the b0 previously (same way we do with the DWI, to create the “b0 pair image”), and now is working fine. The algorithm result is excellent!
Best,