Neurodesk Organisation
Replies: 0 comments 17 replies
|
Dear @c-spelman - We tried reproducing the problem, but haven't been successful yet. We are currently suspecting that this could be a font problem. Could you delete your neurodesk homedirectory and rerun? Does the problem happen again? To delete your neurodesk home directory: Close the neurodesk app. Go to your docker dashboard -> Volumes -> Click the littel trashcan icon under actions: Restart the neurodesk app. Rerun Thank you for reporting this :) I hope we can figure out what causes that. |
|
What's odd is that when I initially preprocessed this participant, the output showed overlapping timestamps for a different task (NumberRight). Then, I re-ran the same participant, using the same script and parameters in Neurodesk, and the output showed a different task with overlapping timestamps (NumberRight looks fine, but now EnglishLeft has the error). |
|
Are you able to share your batch processing code? |
|
Sure, pasted below: mkdir fmriprep /home/jovyan/preprocessing |
|
Not sure why there's a strikethrough appearing through two of the top lines. Screenshot of code below
|
Could it be that something in your BIDS structure is off? so maybe duplicate task files or something like this? |
|
Is there a way to test whether the preprocessed data is fine and this is just a formatting issue? |
|
The actual graphic seems to be generated by this code... https://github.com/nipreps/fmriprep/blob/master/fmriprep/interfaces/confounds.py#L583 calling into this code... https://github.com/nipreps/nireports/blob/main/nireports/reportlets/modality/func.py This feels to me like some global variables are getting mixed up somewhere but I'm not sure if the issue is in fmriprep or nireports or maybe matplotlib. At this stage I strongly doubt it's a local font issue or docker issue since two valid graphics are being rendered on top of each other. I'd suggest we make an upstream issue in https://github.com/nipreps/fmriprep and let them bring it upstream further if needed. |
|
that makes sense. @c-spelman, can you open an issue in the fmriprep repository? |
|
Will do that now. Thank you both for your help! I'll report back with any findings. |
Uh oh!
There was an error while loading. Please reload this page.
Hi, all. I’m using fMRIPrep 23.2.1 (via Neurodesk) to preprocess my fMRI data. I follow the same steps and use the same script for all participants (N = 36). When looking at the .html outputs and doing my Quality Check, I noticed my most recent participants (N = ~5) have overlapping timestamps on the carpetplot for 1-2 tasks. I've added screenshots below with a yellow rectangle around the overlapping values.
Note that the order of tasks does not seem to matter. I have 7 tasks, which populate in the .html in alphabetical order. Various tasks are impacted for each participant; it does not seem to be task-specific, and some participants show this overlap for one task, while others show it for two tasks. I've tried re-running a participant using the most recent version of fMRIprep, and the output resulted in overlapping timestamps for a different task than the one from the output of version 23.2.1.
I'm on a MacOS computer and have updated my Docker memory settings to CPU limit: 12, Memory limit: 24 GB; Swap: 4GB; Disk usage limit: 1TB.
I have confirmation from the scanner facility that nothing changed with the scanner or scanner files around the time I started seeing this error. However, the timing does align with other computer issues I experienced - Docker was alerting me of Malware and forced an update/reinstall (based on online forums, others experienced this as well). I solved the Docker issue by updating/reinstalling, and have un-installed and re-installed Neurodesk since then as well.
I tried re-running an impacted participant after the Docker and Neurodesk re-installs, and the outputs still resulted in overlapping timestamps, but for two different tasks than the previous output.
Any insight or suggestions would be greatly appreciated!
All reactions