Hi michael,
On 30 Jan 2010, at 21:39, Michael Scheel wrote:
> Hi Saad, could you explain a bit more in detail how to proceed.
> When I run probtrackx with and without the --pd option I have the two
> probabilistic tracking distributions - how do I get from them the
> expected path length?
>
> Would it work like this?
> fslmaths corrected_path_distribution.nii.gz -div
> uncorrected_path_distribution.nii.gz pathlength.nii.gz
Yes. because the --pd option outputs the spatial pdf times the
expected distance.
>
>> (make sure you use waypoint and stopping mask to select the tract
>> of interest)
>
>
> Why is that so important? For testing I used a single voxel as seed
> and this seemed to work.
> However what are the reasons for using waypoint and stopping mask?
Because the idea was to plot FA along a specific tract. One way to
isolate a specific tract is to use waypoint/exclusion masks. If you
don't want to plot FA beyond a target region then you can use stopping
masks.
>
>> plot FA etc. as a function of expected distance (it will be a cloud
>> of voxels that you could re-bin according to distance).
>
>
> How would I do this? I suppose there is a simple matlab command?
you can use the path distribution to create a mask of the tract, then
fslmeants to extract FA and expected distance from that mask.
In matlab:
load('fa.txt');
load('dist.txt');
plot(dist,fa,'.');
You can re-bin the distances using hist and calculate mean FA in each
bin.
I can send you some code if you find it hard to do this in matlab.
Cheers,
Saad.
>
> Thanks, Michael
>
>
>
> On 29.01.2010, at 13:20, Martin Kavec wrote:
>
>> Hi Saad,
>>
>> makes a lot of sense. I'll certainly give it a try, to see what it
>> gives in
>> reality. However, this is only applicable to the tracks, which do
>> not branch.
>>
>> cheers,
>>
>> Martin
>>
>> On Friday 29 January 2010 12:10:54 Saad Jbabdi wrote:
>>> Hi
>>> Here is a simpler alternative:
>>>
>>> - run probtrackx with and without --pd to get expected path length
>>> (make sure you use waypoint and stopping mask to select the tract of
>>> interest)
>>> - plot FA etc. as a function of expected distance (it will be a
>>> cloud
>>> of voxels that you could re-bin according to distance).
>>>
>>> The only thing is that when comparing different subjects, you want
>>> to
>>> make sure you take these measurements along the same tracts, and
>>> avoid
>>> noisy data points from alternative routes that are not common across
>>> subjects, but you can solve that by looking only at the overlapping
>>> voxels across subjects in some common space.
>>>
>>> Does this make sense?
>>>
>>> Cheers,
>>> Saad.
>
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|