:)  no problem :) I try to put in very less words.
can you please give me detail overview/procedure , how group analysis & group statistics should be performed, given several individual subject's probtrackx results calculated using 2 approaches (single mask [single track file per mask] &, classification based results [ here with classification I mean, same approach like Prof. Tim behrens did with thalamus ] )

second question, would be how to compare and calculate stats between groups.

Please consider me novice in diffusion data group analysis, and feel free to ask, if above questions aren't comprehensible.

Thank you.

On Mon, Jan 30, 2012 at 3:41 PM, Saad Jbabdi <[log in to unmask]> wrote:


thank you for reply Saad, 

you are right, email was unspecific. sorry for not to be in detail. yes, would like to compare differences in two groups i.e. Male & female, and also within group want to calculate hemispheric differences i.e. Left & Right.  following describes, what I want to compare.

1. Single Mask:
I am taking few seed regions (sub cortical) in both sides of brain for both groups and running them as single mask.

Single mask : Generates a connectivity distribution from a user-specified region of interest.            fdt_paths - a 3D image file containing the output connectivity distribution to the seed mask.( as mentioned at http://www.fmrib.ox.ac.uk/fsl/fdt/fdt_probtrackx.html)

so now I have several fdt_paths for each hemisphere for all subjects.  so, i thought may be I calculate average of them (as u figured out in the command) . and then compare both hemisphere tracks, in both groups.

no need for averaging. just feed the fdt_paths to randomise after concatenating all your subjects (assuming they are all in standard space).
You can test left-right by flipping your images (fslswapdim) and cutting the brain in half (fslmaths -roi)

2. Classification targets :

In second approach, I took whole cortex (Left and right) as a seed region, and target masks (same subcortical regions, which I tracked in above mentioned step). this way I get cortex segmentation after running (find_the_bigggest) in each subject for both hemisphere.

how I should progress with these results (1. single mask 2. classification targets) for grouping & stats, to answer questions like these single mask connectivity distribution are significantly different then other hemisphere in male group or in female or in both. and classification based segmentation results comparison both sides, and to be able to say. like  .. these segmented cortical region is different or bigger or smaller in one hemisphere then other. .
still may be not very specific questions, please write me if something need to explained more.

i still don't understand this bit :)


have a nice weekend,

On Fri, Jan 27, 2012 at 10:53 AM, Saad Jbabdi <[log in to unmask]> wrote:
Hi Vin

Sorry about the delays in responding to this. I have to say that it is not quite clear what you are trying to do. The commands below calculate some sort of average across subjects - why do you need to do that if you want to compare groups of subjects etc.?
In general, you need to decide what it is that you want to compare, depending on the question that is of interest to you. Once you've done that, I am happy to help you figure out how to do the analysis. But from what you've written below, it is not clear what it is you want to do. Sorry...


Saad




Dear FSL Experts,

How one can perfrom DTI group analysis & statistics for two groups & in each group I also want to compare hemispheric differences. ( to see the hemispheric differences and also differences in the two group)  ....for One seed tracking results ( One seed - one side whole cortex - result: fdt_paths.nii)  and for Classification results (Single Seed-- Multiple targets, and at the end seed segmentation using find_the_biggest , results: seeds_* )

I tried following approach:
1) One seed tracking: All subject's resulted track file, I sumed up, and then make a mean of it.
>fslmaths fdt_path_subject_001.nii -add fdt_path_subject002.nii -div 2 avg_track.nii

now thinking to calculate number of streamlines and connectivity value for each subject, and then put these values in Excel/SPSS and do stasts i.e. Mean streamlines, connectivity value, & t-test/anova in both brain hemispher tracks and in 2 group ... does this make sense ?

2) Classification:
    Each individual subject, I am running as suggested in fsl help page, to calculate find_the_biggest.  and in second step, each resulted segment from all    subjects summing it up and averaging like above.
    #individual
    >find_the_biggest seeds_to_*.* biggest.nii
    >mri_binarize --i biggest.nii --match 1 --o s1_1.nii
#group
>fslmaths 1_biggest_subject01.nii -add 1_biggest_subject02.nii -div 2 1_biggest_avg.nii

Should I first sum all the seeds_to_* with same identity, and then run |find_the_biggest| for group. or Should I run find_the biggest for individual and group these results by adding them. 
and same question like above about stats ?

in case any thing need to explain in detail, please ask.
your reply will indeed, enlighten me for DTI group analysis.  :)
or any other suggestion, which helps in comparative group analysis, would be great.

Thank you  :)




--
Saad Jbabdi
University of Oxford, FMRIB Centre

JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466  (fax 717)












--
Saad Jbabdi
University of Oxford, FMRIB Centre

JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466  (fax 717)