On Thu, Jan 9, 2014 at 5:01 PM, Anderson M. Winkler <[log in to unmask]> wrote:

Hi Himanshu,

Please, see below:

On 08/01/2014 11:28, Himanshu Joshi wrote:

Hello Anderson,

Cluster means command "cluster" like in the below example
cluster-i dr_stage3_corrp_tstat3 -t 0.95 -o dr_stage3_corrp_tstat3_index --osize=dr_stage3_corrp_tstat3_size >>dr_stage3_corrp_tstat3_Cordinate_details.txt

And then using dr_stage3_corrp_tstat3_index file for labeling the differential connectivity maps with the help of probability maps (Harvard cortical and subcortical atlas).

Is this method correct?

I think I have a script that automates this task, please here if it's what you need: http://brainder.org/2012/07/30/automatic-atlas-queries-in-fsl

I had lot of confusion regarding dr_stage3_corrp_tstat$x file generated after dual regression.
what is the default level of significance for dr_stage3_corrp_tstat$x...nii.gz.

There is no default. The image contains the corrected 1-p values, ranging between 0 and 1. It's you who apply the significance threshold after you already have the image, not before.

Be careful not to confuse this significance threshold with the cluster-forming threshold, which is given to randomise with the options -c, -C, -F or -S. None of these options is used by default in the dual_regression script. As you are producing these images, I can only assume that you edited the dual_regression to include these options, which is fine (it's line 180 of the script).

Does it shows difference between two group with corrected covariates as implemented through design.mat and design.con file.

Yes, it does. Of course, as long as the design matrix and contrasts are correct for what you want. See here: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM

Can we report the results of this file?

Yes, sure.

more often this file looks as if it contains spurious information that doesn't make sense always(some very scattered activation maps)

Is the design correct? Was the preprocessing good? Many things can cause undesirable results.

I think this is not the desired (p<0.05) file because that doesn't matches the output with the min (0.95) and max (1) values for same file in fslview is seen.

The image contains 1-p, so whatever you want to do with it, you need to take this into account. Threshold it at 0.05 or 0.95 is completely different, so be careful.

All the best,

Anderson

Kindly assist

Thanking you in anticipation

On Wed, Jan 8, 2014 at 3:19 PM, Anderson M. Winkler <[log in to unmask]> wrote:

Hi Him anshu,

I'm not sure if I follow the question. Do you mean you use the command "cluster", or do you mean cluster-level inference in randomise? In any case, no, there isn't a way, nor is it necessary, to specify the alpha (significance level) before running the dual regression, because this is to be done after you have produced the image with p-values, which happens only at stage 3.

Remember that to use 0.05, you have to use 0.95, because the outputs are always 1-p.

Regarding a default level, no, there isn't one, although it's common to use 0.05 for the corrected p-values.

All the best,

Anderson

On 07/01/2014 13:34, Him anshu wrote:

Hello Anderson,
How I was using alpha value p<0.05 is after generation of dual regression maps (dr_stage3 maps), I use cluster to provide significant threshold level.
Is there a phenomenon where we can provide significant threshold before dual regression so that dr_stage3 file would be generated according to our defined threshold level.
By the way what is the default level of significance for dr_stage3_corrp_tstat ...nii.gz.

--

Himanshu Joshi
M.Tech. Cognitive & Neuroscience.
Senior Research Fellow
Ph.D Scholar,
Department of Psychiatry
NIMHANS, Bangalore

http://mbial.weebly.com/himanshu-joshi.html

Himanshu Joshi
M.Tech. Cognitive & Neuroscience.
Senior Research Fellow
Ph.D Scholar,
Department of Psychiatry
NIMHANS, Bangalore

http://mbial.weebly.com/himanshu-joshi.html