Thanks for the reply. It's actually just a normal first level fMRI analysis ran once on subj1 and once on subj1 after flipping the anatomical and BOLD data for subj1. After flipping I ran recon-all on the flipped subject (treating as new subject - so for this sub LH == RH)

Top left images = fMRI for subj1 and subj1_flipped - smoothing and analysis performed on fsaverage_sym

Top right images = fMRI for subj1 and subj1_flipped - smoothing and analysis performed on individual surface (with --self)

The images below are the respective analyses resampled onto the surface in which the analysis was not performed

I expected the output to be symmetrical (because input is subj1 and subj1_flipped).

However, the analysis performed on fsaverage_sym comes out less symmetrcial than when performed on --self

(Resampling each onto the other surface I think shows that it is not a bias introduced during recon-all of subj1_flipped because the output is still more symmetrical when performed on individual surface and resampled onto fsaverage_sym)

From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Douglas Greve <greve@nmr.mgh.harvard.edu>
Sent: 28 October 2016 18:20
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] fsaverage_sym LR-flipchecks

Hi James, this looks like a fairly complicated analysis, and I'm not sure I understand it all. Are the overlays fMRI or thickness results? How did you generate, for example, the top left images?

On 10/26/16 3:15 AM, James Roe wrote:

Hi Freesurfer experts

I have an ultimate goal of performing direct LH v RH comparisons using the fsaverage_sym template, so these pre-analysis steps are aiming to assess the symmetry of the template.

I have performed a first-level fMRI analysis on a subject and am using this to compare the output with the exact same analysis performed on the same subject with LR-flipped data (anatomical, BOLD runs and B0 maps). I then ran recon-all on this flipped subject (treating flipped subject as new subject).

Attached is a screenshot showing the different analysis outputs.

In the top left image, analyses have been performed and smoothed on fsaverage_sym (RH shows LH flipped anat and BOLD data). As you can see, agreement is generally high, but there are also marked differences, most notably in postcentral/precentral regions.

The bottom left shows this output resampled onto the individual surface (for comparison purposes for the next analysis).

The top right image shows the output when analyses were performed on the individual surfaces of the original and flipped subject. Here, agreement seems much higher, also in postcentral/precentral regions (although still not symmetrical). The bottom right image shows this output resampled onto fsaverage_sym, and agreement remains very high.

So it seems that performing the analysis on fsaverage_sym itself may be affecting the expected symmetry of the output. Originally I aimed to perform a comparison of the vectors of B-values in order to prove symmetry, although I am not sure whether this is a viable option anymore. Do you have any advice as to how I could proceed with this?

Finally, it appears that also in the fsaverage_sym template the L/R vertices do not correspond to one another. So will it be possible to flip the analysis values and template in order to perform LH v RH comparisons at the group level?

System: Linux CentOS 6 x86_64 (64b) stable v5.3.0

Thank you for your time

James
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer