Dear conn experts,

I'm trying to run a between subject analysis in conn. I did all the preprocessing (alignment, smoothing..) directly in conn. However, during finalizing the setup (pressing "done") I keep receiving this error message for the 2nd subject (and some others):
"Error using conn_process (line 554)
No suprathreshold voxels in ROI file D:\\Analysen\14SrestDLM17\c3meana001.nii after erosion (original file contained 7025 suprathreshold voxels)"

Interestingly this message doesnt appear for all subjects, only certain ones (yet too many to eliminate them all from the analysis). Apparently, while creating the eroded file: ec3meana001.nii from c3meana001.nii an error occurs because the eroded file looks destroyed (see enclosed). Only the csf-files (c3meana*.nii) seem to be problematic not the wm-files.
When I disable the erosion procedure (EROSION=0) in conn_process.m then the setup finalization runs through. However, this cant be good solution to the problem.
Do u have any clue where the real problem might lie and what I can do against it?

I'd be grateful for any suggestions!
Best, Maxi

Hi Conn experts,

Maxi, I am having a similar error
Error using ==> conn_process at 595
No suprathreshold voxels in ROI file

for some subjects. Has anyone had any luck with this issue since? An older post in 2014 (Use SPM8 new segment instead of normal seg?) suggested that a similar issue/error due to poor segmentation which should have been solved with updated versions.

Thanks,

Kasey

Hi Kasey,

I would suggest to try with the latest release of CONN and SPM12 and the standard CONN preprocessing pipeline. If the problem persists it indicates that the normalization/segmentation step could not find a good match between your structural volume tissue maps and their associated templates. This could be because some image issues in the original structural volume or because the original structural volumes are not "close enough" to standard orientation for SPM to find their proper orientation (if that is the case, typically manually rotating your structural volumes to better match the standard orientation solves this issue).

Hope this helps
Alfonso

Originally posted by Kasey Hemington:

Dear Alfonso and CONN experts,

I'm running into this same issue in CONN v.17. Here is the specific error message I receive when I run my conn batch:

Error using conn_process (line 610)
No suprathreshold voxels in ROI file /nfs/popasd/CoCoA/RiSE/fsl_analysis/fD_0343/anatomy/wc1cbrain.nii (this typically indicates a problem during normalization/segmentation for this subject; please try re-running normalization/segmentation step after manually re-aligning the structural volumes to better match the default template orientation)

Based on this thread, I checked the coregistration for the anatomical volume which looks nice (attached), as well as the grey matter image itself (also attached). My guess is that there is something wrong with the tissue segmentation, and I'm wondering if there are any suggested workarounds for this issue?

Any advice much appreciated!

Cheers,
Jeremy

Dear Jeremy,

That is strange indeed, the grey matter image and normalization results look perfectly fine. The error indicates that no voxels in your gray matter image have posterior probability values above 0.5, which is somewhat strange judging from the quality of the displayed mask. Could you please, within the SPM display, right-click on the image and select 'display intensities', and double-check whether in fact it appears that all values are below 0.5? Alternatively please mail me the wc1cbrain.nii image for the FD_0343 subject and I will take a quick look to see if that helps me figure out what might be the issue

Thanks
Alfonso
Originally posted by Jeremy Hogeveen:

Alfonso,

I actually managed to rectify the issue offline so further troubleshooting will not be needed. Specifically, i re-ran the preprocessing pipeline after manual re-orientation and that seems to have rectified the issue. Thanks so much for your time and consideration nonetheless!

Cheers,
Jeremy