Hello,

I am running into an issue where my functional data does not appear to be properly registered when I look at the data in the denoising tab. 

I originally loaded in my data, ran preprocessing steps including direct segmentation and registration, and everything looked ok. I started running denoising but I had to quit and shut off my machine because of maintenance scheduled power outage in my building. 

When I reloaded the data, the functional data appeared to not be correctly registered in the denoising tab. Is there something I need to do to get conn to recognize the computed registrations?

Best,
Harrison Fisher
Research Assistant
Napadow Lab

Hi Alfonso, 

Bumping this because I'm still dealing with this problem and not sure what to do about it. I've attached a screenshot of what it looks like at the Denoising step. The overlaid color-coded functional activation images at the First Level Analysis step also are shifted off, but differently for each subject. 

I used the default processing pipeline (direct normalization). The Functional files loaded have the swu* prefix.

Any advice is appreciated. Thanks!

Best,
- Harris

Hi Harris,

Looking at the underlay picture in the snapshot that you sent, it looks like the images are not skull stripped. If the default pipeline ran completely, the resulting structural image would be wc0*.nii. Is it possible that your functional data got loaded into Conn (that's the reason you are seeing the swu* prefix in the functional file names) but that the structural image did not get loaded for some strange reason (perhaps the shutdown you mentioned happened in the middle of the importing step)? Can you go to the Basic tab and check if the structural images have the wc0* prefix?


Hope that helps


Best
Pravesh
Originally posted by Harrison Fisher:

Hi Pravesh,

Yup that was it! I somehow reloaded the original MPRAGE images and not the wc0* prefix images. It all looks ok now. Would this affect the denoising or first level analysis at all? Or would only the displays / visual overlays be different?


Best,
- Harris

Hi Harris,

Glad that the problem got resolved. As far as I know, the structural images are not really used for any further steps (i.e. no processing happens based on the structural images). Denoising steps (band pass filtering, regressing of nuisance covariates etc.) are applied only to functional images. Same holds for first level where correlation (or any other metric that you would like calculated) is calculated from functional image. So, I do not think it would matter really. Since you have already loaded the correct structural images into Conn, Conn should pick them up (if at all they are needed) just fine since the conn project variable would get updated once you save the project.


Best
Pravesh

Originally posted by Harrison Fisher:

Hi Harris,

Just to be extra-safe I would also double-check whether your Gray/White/CSF ROIs are correctly defined (i.e. they point to wc1*, wc2*, and wc3* files, respectively). If that is not the case, then the BOLD timeseries associated with the gray/white/CSF areas may have been extracted from an incorrect location so, in that case, I would suggest to first associate those ROIs with the proper files and then re-run the "ROI extraction" part of Setup step as well as the entire Denoising step (this is just to make sure that the aCompCor procedure is using the correct BOLD timeseries).

Hope this helps
Alfonso 

Originally posted by Pravesh Parekh: