Question on Validity/Appropriateness of Proposed Method

[Sunderland Baker]

Hi there!

I am currently working on a project where we use whole-brain correlational tractography using QA and GFA to identify differences between two groups. Then, we proposed running post-hoc, ROI-wise analyses just within those regions of significance using other DTI metrics to try to get a better idea as to what biologically might be happening to the white matter.

Is this ROI-wise approach logical, or is it an issue that I am converting from fixel-wise to voxel-wise. What I had been doing was creating a mask of the significant results, binarizing it, and using that as a mask to restrict the tractography. However, I have tried multiple such approaches (e.g., ROI, ROA, end, termination, etc.) after going through the tutorials to familiarize myself with what they are doing, but it does not seem to be capturing just this region of significance.

Is a different approach, say, tract-based spatial statistics, more appropriate for these post-hoc analyses we are proposing? Thank you for your help.

An additional question we had was how the tractography is computing a general tractogram (i.e., the .tt.gz file) across an entire sample? What approach is used to threshold these images to gain the "most likely signature" across a sample?

Best,

Sunderland Baker

Department of Biobehavioral Health, Penn State University

[Frank Yeh]

Hi Sunderland,

As to your questions:

> Is this ROI-wise approach logical, or is it an issue that I am converting from fixel-wise to voxel-wise.

It is logical and have been done many times in other studies.

> What I had been doing was creating a mask of the significant results, binarizing it,

We might do the same until this step.

> and using that as a mask to restrict the tractography.

This may be the point where it may not work out.
I would just get region statistics and correlate the statistics
with study variables.

>
> Is a different approach, say, tract-based spatial statistics, more appropriate for these post-hoc analyses we are proposing? Thank you for your help.

TBSS is a different framework, and won't be able to use
correlational tractography results because the findings are limited on
skeleton.

>
> An additional question we had was how the tractography is computing a general tractogram (i.e., the .tt.gz file) across an entire sample? What approach is used to threshold these images to gain the "most likely signature" across a sample?

I recommend taking a look at the workshop videos available at
practicum.labsolver.org. They provide comprehensive insights into
conventional, differential, and correlational tractography.

It's completely understandable to find correlational tractography
challenging to start with. Its concepts are actually built upon
conventional and differential tractography, so understanding those
first can often make it much clearer.

Best,
Frank

>
> Best,
> Sunderland Baker
> Department of Biobehavioral Health, Penn State University
>

> --
> You received this message because you are subscribed to the Google Groups "DSI Studio" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/dsi-studio/321d1459-7cb0-41fa-9431-1d44cfad0fa0n%40googlegroups.com.