Agnieszka Burzynska wrote:
Dear Doug, Thank you a lot for your reply!
Yes, I did use reg-feat2anat and registered example_func to the subject's anatomical. As far as I understand it, then the registration matrix can be used for a contrast of interest.
That's great news that the software version will not be a problem.
I will use the copes, thank you.
My cope images have a range of, for example, -500 to 500 with the peak at 0. When I view them on the subject anatomical in the freesurfer
(with command tkmedit 1109 orig.mgz lh.white -aux brain.mgz -overlay ./run1.feat/stats/cope1.nii.gz -overlay-reg ./run1.feat/reg/freesurfer/anat2exf.register.dat -segmentation aparc+aseg.mgz -fthresh 2.3 -fmax 4.3 ),
the -fthresh 2.3 and -fmax 4.3 are not suitable. However, when I try to adjust the threshold in View/Configure/Functional Overlay I get this error message:
Invalid values. Tkmedit couldn't set the threshold to the values you specfied. Please make sure the minimum is less than or equal to the midpoint, and the slope is greater than 0.
What are you trying to adjust them to?
I am not sure how to deal with this.
Now the last point:
- I will want to create a fMRI-based ROI and then investigate the
thickness especially within this region. i know I can extract mean ROI values and use other statistical software, but how can I perform vertex-wise analysis within this ROI? I assume qdec does not allow this, so only the DODDS analysis is possible.
Which stat map do you want to derive the ROI from? The group map?
Yes, my first idea was to create a conjunction image across all subjects, specify the ROI on the fsaverage, then map this ROI back to subjects' individual surfaces. And then see whether across subjects, thickness is related to fmri data within this cortical patch.
If you are going to create a conjunction, then you should use the zstats. But then you have another problem in that the zstats are going to be very low because no smoothing was done. But if you're happy with the maps, then go for it. Alternatively, you can do a group analysis of the copes and then threshold the sig maps. In either case, the map will be in the fsaverage space. You can create a label in tksurfer. Then run mri_preproc to get all of your subjects thickness files into one file in fsaverage space. Then run mri_segstats using the --slabel and --avgwf options with the thickness file as input. The avgwf file will be a list of thicknesses for each subject averaged over the ROI.
doug
Thank you again for your help! Cheers, Aga
On 5/10/11 4:39 PM, "Douglas N Greve" greve@nmr.mgh.harvard.edu wrote:
Hi Aga, comments below
Burzynska, Aga wrote:
Dear all, I just wanted to double-check if I am doing the right thing.
I am transferring fMRI data from FSL to the surface in Freesurfer to test whether cortical thickness is related to BOLD signal change.
I used the steps described here: http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/FslFeatFreeSurfer
I did the following:
- I carried out Freesurfer reconstruction for all subjects
- I performed 1st level analysis for fMRI data in FSL, separately for
each run, without smoothing 3) Then I obtained good registration of FEAT example_func of each run to the subject's anatomical using FLIRT. This I have done already for a specific contrast of interest.
Did you not use reg-feat2anat for this step? Did you register the z-map itself? If so, that is not the right thing to do, it should be the example_func.
The above steps have been performed with Freesurfer version earlier than 5.0. Will there be a problme if I continue the steps below with the newest stable version?
This will not be a problem.
Now I plan: 4) Combine the 3 runs/subject on the surface by averaging them with smoothing. As this is within-subject, my guess is that 5mm will be the best.
Sounds good. Make sure to use the copes and not the zstats
- Concatenate the images for all subjects to perform analyses
I am not sure about the following points:
- what are the units of the zstat images transferred onto the surface?
Is it still the logarithmic scale as for thickness analyses?
I'm not sure what you mean. As I mention above, you should use the copes and not the zstats for the next higher level analysis.
- I will want to create a fMRI-based ROI and then investigate the
thickness especially within this region. i know I can extract mean ROI values and use other statistical software, but how can I perform vertex-wise analysis within this ROI? I assume qdec does not allow this, so only the DODDS analysis is possible.
Which stat map do you want to derive the ROI from? The group map?
doug
Thank you a lot for your support! Best, Aga
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer