Hi Seda,

Yes, as you add each participant, definitely your DTI values do change. TBSS uses the whole participant to create a skeleton, and then it's easy to extract the participant database.

Is the participant you're adding to the study is the follow-up of the participant?

If not, you need to collect all the participant, then run TBSS to get the skeleton--stats.

Do follow this link, its is crisp and clear

http://jeromemallershandbookofstructuralbrainmrianalysis.yolasite.com/

I hope this helps

Cheers

Sam

Hi Samson,

Actually, we are collecting new participants almost every week. So, it is not a follow up, it is definitely a new brain data.

Thank you for the answer. I checked the link you sent. It is very helpful. But, as I understand, there is no such way to get FA and MD before TBSS, then.

Hi James,

Thank you for this useful answer forum. I am posting my question there, too.

Best.

Hi Seda,

one way to extract FA and MD "before" TBSS would be to not use TBSS. As you correctly state, TBSS is based on a group specific template and therefore the recreation of that template every time new data arrives would change the subject specific values. Here you have the inherent disadvantage of group templates: constant re-calculation and changing properties when the input data constantly changes.

What I would use here are two things:

Proper calculation of a normalization of the DWI data into MNI space (FSL FLIRT+FNIRT, ANTs SyN). Perhaps via a high res T1w anatomical image. Or, as TBSS is doing it, using the subject specific FA image and utilizing the FA MNI Template that is included in current FSL releases. Of course, that normalization wouldn't be that precise as with a group template, but it's always a trade-off.

Second, find some ROIs in MNI space that fit your expectations. There are some good atlases out there. Use either the inverse of the above transformations to MNI space or transform your FA/MD images to MNI space. Latter will introduce interpolation artifacts such as smoothing, the first option could reduce your MNI space ROIs to tiny ROIs in subject space, if the spatial resolution of your DWI data is coarse.

Either way, use fslstats to extract your values using the ROI in the correct space as a mask. That's it.

But, your data will need complete reprocessing in regards to image registration…

Best,

Martin

Thank you very much Martin!

It is really helpful. I will try those suggestions.

Best,

Seda