Hello!

I had a question about analyses completed with conn using the surface based analysis pipeline (Freesurfer structural images, same subject-space functional images). In the second level analysis, how could I export/find the raw (not thresholded) data from my contrast? E.g. a map of the t or p values resulting from a between-groups contrast [1 -1] which I could then correct and threshold myself? Alternatively, in the results explorer, it is only possible to view the positive or negative contrast separately and export these separately--is there a way to view them concurrently? 

Thank you in advance! This toolbox has been incredibly useful!

Hi Shipra,

The files created by the 'export mask' button will contain the actual T/F statistics for any suprathreshold vertex so you should be able to simply use a most-liberal threhsold (e.g. p<1) and have those files include the stats for your between-group contrast for every vertex.

And regarding the two-sided display, sorry this option is not yet included in the gui, thank you for the suggestion and I will add this option to the next release of the toolbox.

Best
Alfonso
Originally posted by Shipra Kanjlia:

Thank you so much!

Hello again--I am trying to do connectivity analyses by aligning participants' language dominant hemispheres. I've tried doing this a multitude of ways but haven't had much luck. (E.g. I've tried, outside of conn, to map each subject's language dominant hemisphere data to the left volume--however this is going from surface data to volume which causes me to lose white matter and csf, which is what I'm assuming interfered with the conn analyses running properly). I thought I could try aligning each subject's language dominant hemisphere in the first level results (of surface based analyses I ran in conn that worked beautifully) but have had some trouble doing that as well--namely the fsl and hcp workbench commands I've tried haven't been able to successfully align subjects' language dominant hemispheres (i.e. take the data from their lang dom hem and map it to the left hemisphere)

Is there anything you would suggest to try that might work to do language dominant hemisphere analyses?

Hi Shipra,

Unfortunately, as far as I can tell, the best approach would require you to re-run your analyses almost from the beginning, mainly because the matching between left and right hemisphere surface vertices is far from straightforward (if you simply switch left- and right- hemisphere surface data you would end up with lateral activations being projected to medial areas and viceversa) so performing this alignment on the first-level results is rather complicated (it can be done, it just requires a lot of coding). So, instead, I would suggest to simply flip the original/raw functional and anatomical data of the subjects with righ-hemisphere dominance, and then proceed normally for the rest of analysis steps. You could do this, for example, using the following steps:

1) first enter into your CONN project your raw structural and functional volumes for each subject
2) then go to Setup.Structurals and click on 'structural tools', select the 'manual reorient' option, uncheck the 'All subjects' box and select only the subjects with right-hemisphere dominance, click 'Ok' and then select the option that reads 'non-rigid reflection of x-axis'. That will flip the x-axis of your structural data (typically the left-right direction, change accordingly to a different axis if your data is not in standard orientation)
3) after this, repeat the same step for your functional data (in Setup.functionals, select 'functional tools', 'manual reorient', etc.).
4) then run FreeSurfer using the flipped structural volumes for all of the right-dominant subjets (this is necessary to make sure that the FreeSurfer surfaces are appropriately coregistered to the new/flipped structural volumes)
5) after you have done all of this you can run the standard preprocessing pipeline (realignment, coregistration, segmentation, etc.) in CONN and continue with the rest of the connectivity analyses in surface space normally

If this is not cost-effective for your project and you would prefer to attempt to perform the alignment on the first-level results, let me know and I will try to guide you through some of the issues there. Alternatively, some of the above steps could also be skipped with some minor modifications (e.g. you could keep the realigned, slice-corrected functionals in (1) and remove those steps from (5); you could potentially skip (4) by manually editing the surface info in each subject FreeSurfer folder, etc.)

Hope this helps
Alfonso
 

Originally posted by Shipra Kanjlia:

Hi Alfonso,

This has been incredibly helpful! Thank you! I am experiencing one problem however--when I manually orient the functional data, it looks like the first scan is successfully being flipped but not the last scan (according to the conn manual, the left image is the first scan and the right image is the right scan) (I have a total of 240 volumes during the resting state session). I've attached an image of this. I think this is causing a huge difference in distribution of voxel to voxel r values (displayed during the denoising) between flipped and non-flipped subjects (image of this also attached). Is there a way to make sure all of the scans are flipped? Thank you again!

Hi Shipra,

Thank you for reporting this, and sorry for the late reply. I looked into this issue in more detail and it seems that it was a combination of a bug in CONN and another one in SPM that produced this effect (and complicated fixing this issue a bit). In any way, the attached patch should fix this (even if the SPM version is not patched, which will be in their next SPM12 release).

To fix your subjects' data, please use the following steps:

1) install the attached patch (copy that file to the conn distribution folder overwriting the file with the same name there)

2) for each right-handed subject raw structural file file [...anatreg.nii], delete the associated [...anatreg.mat] file in the same folder (this will undo any reorientation applied to these structural files reverting them to their original orientation)

3) apply again the left-right flipping of the right-handed subjects [...anatreg.nii] structural files (e.g. in conn select 'structural tools', 'manual reorient', etc.)
Let me know if everything works as expected now

Best
Alfonso
ps. if you want to fix the related SPM issue before the proper patch is released you may do so simply by modifying spm_get_space.m line 30 to read

instead of



Originally posted by Shipra Kanjlia: