Hi Wolf,

What I want to do at the end is a seed-based-connectivity analysis between two groups....so...maybe I should register the roi to the individual subject space, shouldn't? Sorry I did not tell you this before.

By the other hand, I have tryed and it appears and error that says: lacking argument for -vox-

Thank you a lot
Rosalia.

2013/7/21 wolf zinke <[log in to unmask]>

ah i see. So you want to use the ROI to reduce the number of voxels for your statistical tests and not to derive values from these voxels for further analysis. If you use the ROI with randomise on the group statistics, there is no need to register a ROI to the individual subject space. Just create it in you 4mm MNI space and use it directly with randomise.

hope this helps,
wolf

On 21/07/13 20:25, Rosalia Dacosta Aguayo wrote:

Hi  Wolf,

Sorry if I have not explained this properly.

1. Draw the ROI in the 4mm MNI space....make the mask....
2. Registration of this ROI to the subject space (native) that has 4mm of resolution
3. Registration of ROI of every subject space to MNI space using standard2examle_func.mat (FLIRT)
4. Make analysis (randomise)
5. In the case I get corrected results...resample the results to a 2mm MNI space template.

Is clearer now?

Thanks a lot for your help,
Rosalia.

2013/7/21 wolf zinke <[log in to unmask]>

Hi,

I can't completely follow. Do you want to use the ROI in the 4mm MNI space, or in the 4mm subject space that is different from your 4mm MNI space?

In the end, the decision where to create the ROI depends on whether you want to ensure that the ROIs for each subject have the same number of voxels (then you have to transform the center coordinate into the subject space and create the sphere there) or whether the ROI should represent the same fraction of the template space (then you create the ROI in MNI space and transform it into the subject space). For the latter, it does not really matter where to create the ROI, because it will be transformed anyway. Thus, you might here be better off by creating the ROI in the 2mm space..

Maybe others have here more grounded opinions about this issue.

cheers,
wolf

On 21/07/13 19:13, Rosalia Dacosta Aguayo wrote:

Hi,

I did it in a template of 4mm of resolution (MNI space). But anyway, I can do it in a 2mm because at the end of all these steps I am going to resample the 4mm image to 2mm image (MNI)...so, what do you recommend me? in 2mm directly or in 4mm. But I am in the same place.....4mm to 2mm 2mm to 4mm.

My subject's fmri images are in 4mm resolution...so, for registering the rois to the subject's 4mm resolution and then make analysis and finally report the results in a 2mm resolution...because it is more anatomically clear....

So?...I am confused about the way to follow.....arfssss

Kind regards,
Rosalia.

2013/7/21 wolf zinke <[log in to unmask]>

Hi,

In which image space did you specified the center coordinates? If the coordinates are specified in the 2mm MNI space, then it might be good to transform only this coordinate into you 4mm space using the std2imgcoord with the -xfm option and create the sphere in the 4mm subject space in order to ensure the ROI size. If you transform the rois from your 2mm space into the 4mm space, you run again into interpolation and thresholding issues that might  result in spheres of slightly different sizes for each subject (and slightly means, for a binarized ROI in a 4mm space, differences of +/- 4mm).

Once you created for each subject the spherical ROI, you can transform them for display purposes back into the 2mm space, and sum them up in order to visualize the overlap of all individual ROIs.

hope this helps,
wolf

On 21/07/13 18:41, Rosalia Dacosta Aguayo wrote:

Sorry,

As I am working with fmri data resampled at 4mm....maybe should be better use a MNI template of 4mm....but I am worried about when I want to register this rois back to every subject image and then ApplyXFM to a 2mm template...Do you know if I am going to have any problem with this?

Thanks again!
Rosalia.

2013/7/21 Rosalia Dacosta Aguayo <[log in to unmask]>

Hi Wolf!

Thanks a lot for your helping on Sunday :)

So, following your advices and the answered mail of Mark Jenkinson I am going to do as follows:

echo "22 42 13" | std2imgcoord -img $FSLDIR/data/standard/MNI152_T1_2mm -std $FSLDIR/data/standard/MNI152_T1_1mm -vox -

fslmaths $FSLDIR/data/standard/MNI152_T1_2mm -mul 0 -add 1 -roi A 1 B 1 C 1 0 1 -kernel sphere 10 -dilF medial_frontal_cortex_roi

fslmaths medial_frontal_cortex_roi  -thr 0.5 -bin medial_frontal_cortex_roi_mask


That's all!!!  (I hope...)

Thank you a lot for your time and your helping.
Kind regards,
Rosalia.

Note: I hope you have finished your PhD!!!