Hi Davide,
1. You can run tractography to find the most likely path of the WM tract
connecting the ROIs. In FSL this is available as probtrackX
(https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FDT/UserGuide#PROBTRACKX_-_probabilistic_tracking_with_crossing_fibres).
This will give you for every subject a probability map of where the
fibers connecting the ROIs might run, which you can use to calculate an
average FA for that subject.
2. You can provide ROIs in either standard or subject space to
probtrackX (see
http://fsl.fmrib.ox.ac.uk/fslcourse/lectures/practicals/fdt2/index.html#tractography
for some examples). Your choice here will depend mainly on which space
you have defined the ROIs in.
3. I am not sure what you mean with a "linear connection". If you mean
that there might not be any fibers connecting your ROIs, it is important
to note that tractography (and especially probabilistic tractography
like that used by probtrackX) tends to have a lot of false positives. So
even if there are no actual fibers, probtrackX is still likely to give
you some potential paths. Determining whether there is a connection is
impossible based on diffusion data alone, so you will have to search the
literature for prior evidence for or against fibers connecting the ROIs.
If there are no such fibers, then I do not think it is meaningful talk
about "the FA values between two ROIs regions", so you will have to
rethink on what you want to extract from this data.
Best wishes,
Michiel
On 26/05/2018 16:50, Davide Momi wrote:
> Dear FSL users,
>
> I have some questions regarding DTI analysis. Specifically I’ve estimated the FA map of one subject. Now I want to extract the FA values between two ROIs regions (e.g. between Angular gyrus left and right).
>
> 1. Which analysis do you suggest to run in order to find FA values which connect this two regions?
>
> 2. In which space do you suggest run this analysis (subject space or MNI)
>
> 3. Also I’m not sure that there is a linear connection between this two regions. In this case what do you suggest to do?
>
> Thanks
>
> Davide
|