Hi Prerona,

 

You have a big sample size.  That’s always great.

 

Here are my suggestions.

 

Example for your EV #1:

 

File:  ev1.con for your EV #1 (ev1.con – you can make a text file as follows)

 

/ContrastName1 “ev1pos”

/ContrastName2 “ev1neg”

/NumWaves      4

/NumContrasts 2

 

/Matrix

1              0

-1            0

 

 

ev1.mat

/NumWaves      4

/NumPoints       102

 

/Matrix

Here you can have four demeaned columns (first column – what you’re interested –EV1, second-Age, third-gender, fourth-handness)

 

ev1.con and ev1.mat is ready.

 

Now you need to get your mask for the specific tract you’re interested.  For example, you are interested in Superior longitudinal fasciculus (temporal part) L.  Use the following command to get the mask ready:

 

fslmaths $FSLDIR/data/atlases/JHU/JHU-ICBM-tracts-maxprob-thr0-1mm.nii.gz –thr 19 –uthr 19 AF_L

 

19 is the index value for that specific tract.

 

$FSLDIR/data/atlases/JHU-tracts.xml – you can find more values in this file.

 

1              Anterior thalamic radiation L

2              Anterior thalamic radiation R

3              Corticospinal tract L

4              Corticospinal tract R

11           Inferior fronto-occipital fasciculus L

12           Inferior fronto-occipital fasciculus R

13           Inferior longitudinal fasciculus L

14           Inferior longitudinal fasciculus R

15           Superior longitudinal fasciculus L

16           Superior longitudinal fasciculus R

19           Superior longitudinal fasciculus (temporal part) L

20           Superior longitudinal fasciculus (temporal part) R

 

etc.

 

At last, you can do your non-parametric correlation test.

 

randomise -i all_FA_skeletonised -o tbss_AF_L -m AF_L -d ev1.mat -t ev1.con -n 500 --T2 –V –D

explanation:

-i: input volume

-o: output volume

-m: your mask – here you can use your specific prior hypothesized tract  (by the way, you can further narrow it by mask the tract using mean_FA_skeleton_mask)

-d: .mat design matrix

-t: .con contrast definition 

-n: default: 5000, you can use 500 to see the preliminary results.

Hope these are helpful to you.

Merry Christmas!

Yingying

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of s0675204
Sent: Wednesday, December 19, 2012 1:14 PM
To: [log in to unmask]
Subject: [FSL] TBSS with ROI

 

Hello

I am trying to run a 102 subject, one group, correlation analysis using TBSS. I have 2 EV's of interest and 3 co-variates of no interest (age, sex, handedness). I want to look at FA variation with my EV of interest only within particular tract for which I have a a-priori hypothesis.

Since I have 102 subjects, hand drawing would be very time consuming.
I would be extremely grateful if anyone can help me with this.

best wishes,

Prerona

On 18 December 2012 01:07, Stephen Smith <[log in to unmask]> wrote:

When you setup your account on jiscmail you selected "digest" option - meaning that you just get a summary of all messages every now and then - I have changed your setting to "regular" so you get separate emails.

 

Steve.

 

 

On 17 Dec 2012, at 18:59, Prerona Mukherjee wrote:

Thanks so much!

PS: For some reason the mailing list replies never come to my mail box - is this as it is supposed to be?

regards

Prerona

 

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------