As request, I'm including the previous mail on this topic.
Subject:
Re: Concatenate files and randomise
From:Stefano <[log in to unmask]>
Reply-To:
FSL - FMRIB's Software Library <[log in to unmask]>
Date: Thu, 27 Feb 2014 10:29:39 +0000
Dear Mark,
the images are well aligned but, after warning, the computation is stopped.
Thank you very much,
Hi,
If you've checked the images and they look well aligned, and the anatomical labels are still in the correct places, then you can ignore this warning.
All the best,
Mark
Hi,
as known by previous mail in this topic I'd like to performed a voxel-vased statistic on probtractx output.
To improve the co-registration between probtrackx output I'm using fslroi.
fslroi <input> <output> <xmin> <xsize> <ymin> <ysize> <zmin> <zsize>
In this way, the probtrackx outputs result well aligned for voxel-based statistic.
Before to performed randomize, I used fslmerge to concatenated all files.
When I run that after fslroi I have this error (it occur only after fsl roi correction):
WARNING:: Inconsistent orientations for individual images when attempting to merge.
Merge will use voxel-based orientation which is probably incorrect - *PLEASE CHECK*!
The fslroi seem be well aligned. fslinfo produce these info for all files:
data_type FLOAT32
dim1 182
dim2 218
dim3 182
dim4 1
datatype 16
pixdim1 1.0000000000
pixdim2 1.0000000000
pixdim3 1.0000000000
pixdim4 1.0000000000
Difference were found only for cal_max.
How can I resolve?
Thank you very much,
Stefano
Subject: Re: Concatenate files and randomise
From: "Anderson M. Winkler" <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Mon, 17 Feb 2014 17:10:45 +0000
Content-Type:
Hi Stefano,
As long as each voxel (i.e., a position in space) contains a measurement
of the same thing across subjects, it's ok to use randomise. Of course,
it's up to you to interpret the results later, but from the stats
viewpoint it's ok.
I had thought you'd like compare the paths themselves, which vary for
each subject, and so, a single voxel across all wouldn't contain the
same thing: some would be at the center of the tract, some in the
periphery, some more anterior, some more lateral, and some voxels would
be entirely outside the putative tract. Then randomise would not give
meaningful results (although without any error message).
All the best,
Anderson
Am 17.02.14 16:19, schrieb Stefano:
> Thank you very much.
> Please look the figure 2A of the paper Nair et al 2013 Impaired thalamocortical connectivity in autism spectrum disorder: a study of functional and anatomical connectivity.
> It is more explicative to understand my question.
>
> Stefano
Hi Andreas.
In that case, it's fine to compare the counts for the seeds themselves,
as opposed to the tracts. And fslmerge is just ok then.
Thanks!
All the best,
Anderson
Hi,
I think Stefano uses the voxelwise counts of probtrackx streamlines that
make it to specific targets from predefined seeds in standard space which should be aligned.
So even though there may be other issues to consider, alignement should
not be the major one.
Cheers,
Andreas
Am 17.02.14 16:44 schrieb "Anderson M. Winkler" unter
<[log in to unmask]>:
Hi Stefano,
I don't think it's wise to use randomise in the outputs of probtrackx,
because the tracts aren't aligned to each other. If you'd like to
compare the measurements from these tracts, perhaps you could use them
as masks to extract some other measurement, e.g., FA, etc. You'd have
then a single measurement per subject, which you could analyse in any
statistical package.
All the best,
Anderson
Subject: Re: Concatenate files and randomise
From: Stefano <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Mon, 17 Feb 2014 11:42:07 +0000
Hi again,
I'm thinking to use fslmerge to concatenated the output of probtrackx (e.g. seed_to_frontal.nii.gz) of each subject.
fslmerge -t $SUBJ01/seed_to_frontal.nii.gz $SUBJ02/seed_to_frontal.nii.gz ... 4D_final_output
Thus, randomise could be performed for within and between groups comparisons.
What do you think about it?
Please, give me a feedback.
Thanks in advance and best regards,
Stefano
Subject: Concatenate files and randomise
From: Stefano <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Sun, 16 Feb 2014 13:45:21 +0000
Hi list,
I have carried out hard segmentation of thalami.
Now, having 3 groups, I'd like to performed within and between groups probabilistic tractography comparisons for probtrackx output (e.g. seed_Frontal_cortex.nii.gz).
I'm thinking to use randomise. How I can concatenated in a one file the seed_Frontal_cortex.nii.gz of all subjects?
Stefano
|