Open araikes opened 5 years ago
Hi @araikes yes, you used current fmriprep version 1.4x with different output-spaces. we currently updated the xcpengine as well to match the current fmriprep outputs with different template. you can pull for new docker/singularity image.
Hi @a3sha2, As I noted in my post, I have the current version from Dockerhub. I just double checked this. I have both the 1.0 tag and the latest tag and they are up to date. Do I need to pull straight from the repo and build the Docker image?
Assuming that it was a template spaces issue, I would expect it to be looking in the correct file location but have trouble identifying the correct MNI file to use. Instead, it's looking for the correct string *MNI152NLin2009cAsym*
but it's not looking in the correct location for a multi-session study.
Yes, we have not change the volume number but it has been updated. You can pull for new one. it will work fine
Hi @a3sha2, Please forgive me. I feel like I'm not communicating this clearly.
I have the current version of XCP Engine. I pulled it yesterday and tried again this morning before I responded earlier, just to confirm. The version on which I'm having this issue is the most current version.
thanks @araikes pls pull for new docker image
Pulled it and it appears to be working as expected. I'll keep an eye on it and let you know.
I am now running into a new issue, though I can't fully tell if it is problematic. Steps up until fcon work just fine. However, the following error is thrown:
Current processing step:
@1 Functional connectome: HarvardOxford
····································································
· @1.2 Mapping network to image space
· @1.2.2
- - -
>> fslstats /xcpEngine/atlas/HarvardOxford/HarvardOxfordMNI.nii.gz -R
- - -
>> 8058.000000<=1
Error in parse_args(OptionParser(option_list = option_list)) :
Error in getopt(spec = spec, opt = args) : flag "e" requires an argument
Execution halted
- - -
>> rln /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/sub-BLMTBI104_ses-bline_atlas/sub-BLMTBI104_ses-bline_HarvardOxford.nii.gz /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford.nii.gz
- - -
>> 8-5-1
- - -
>> 8-1
· @1.3 Computing network timeseries
- - -
>> rm -f /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford_ts.1D
- - -
>> ${XCPEDIR}/utils/roi2ts.R -i /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/sub-BLMTBI104_ses-bline.nii.gz -r /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/sub-BLMTBI104_ses-bline_atlas/sub-BLMTBI104_ses-bline_HarvardOxford.nii.gz -l /xcpEngine/atlas/HarvardOxford/HarvardOxfordNodeIndex.1D
· @1.4 Computing adjacency matrix
- - -
>> rm -f /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford_network.txt
- - -
>> rm -f /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford.net
- - -
>> rm -f /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford_missing.txt
- - -
>> ${XCPEDIR}/utils/ts2adjmat.R -t /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford_ts.1D
- - -
>> ${XCPEDIR}/utils/adjmat2pajek.R -a /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford_network.txt -t N
· @1.5 Determining node coverage
- - -
>> ${XCPEDIR}/utils/missingIdx.R -i /data/xcpengine/mni-aligned/sub-BLMTBI104/ses-bline/fcon/HarvardOxford/sub-BLMTBI104_ses-bline_HarvardOxford_network.txt
This happens for all atlases. Is this functioning as intended (spurious error) or is it skipping something and jumping to computing subroutine 1.3?
Hi @araikes
There is subroutine 1.2.3 or 1.2.6 that is compulsory which look like it doesn't run that is why there was error. that subroutine is necessary when your bold is not in MNI (or template) space. it is missing some transformation files from fmriprep ouput. You have no problem if your bold is in MNI space but it will be nice to know what is going on. Can you share with us the full log file and the _space.json
Also try to rerun with prestats_rerun[1]=1
Hi @a3sha2,
So I went back and did a clean processing and it appears to work. That particular error only shows up if I have to restart/rerun due to failure (we had some network failures leading to early processing termination).
However, if I start from scratch or presumably if I do prestats_rerun[1]=1
, it shouldn't throw an issue.
setting prestats_rerun[1]=1
will ensures that the inputs are the same with previous and read all the transformation files needed. We will make sure it is default in all the design file #221
Hi @a3sha2, Everything appears to be working well for the anatomical processing stream now.
I do have one further question: In fMRIPrep, one of the preprocessing steps is N4BiasCorrection. XCP Engine appears to also be running N4 correction. Is this double correction potentially problematic?
It may be problematic honestly. Running N4 twice or many time can increase or decrease voxel intensity in some regions. The best is to use use raw image ( from bids) as input. Look at difference between N4 (once) and N4(twice)
I figured that would be the case. The documentation should probably be revised to reflect using the original images rather than the fMRIPrep'd ones. Currently, users are instructed to use the post-fMRIPrep files prior to running one of the Anatomical pipelines (see image below).
Also, from the way that the Functional pipelines currently work, would using the original BIDS images pose any problems if the anatomical stream was run first to pipe into the functional stream? I can see it may pose issues if longitudinal/multi-session T1s are merged in fMRIPrep. Maybe post-fMRIPrep T1s need a flag/option to skip N4.
thank you very much. we will make correction
Hi @a3sha2, I'm just getting back around to this. I looked at the update to the documentation (#228), and that file isn't part of the fMRIPrep output tree. If the intent is to use the original T1w, then the listed path shouldn't be under the fMRIPrep derivative.
Additionally, I tried using the raw T1w image and it does not work correctly. My cohort file uses the original T1 in my BIDS nifti directory and the following is the sub-xxx_ses-xxx_struct_report.svg
generated:
pls share me the logs files pls
Describe the bug I was attempting to run XCP Engine on a multi-session dataset. My target was the MNI152NLin2009 space. It appears that XCP (at least as far as mine is running) is looking in the wrong spot for anatomical data in a multi-session fMRIPrep tree. I am aware of the potential issues with the multiple template spaces. However, XCP Engine isn't even looking in the correct folder for the anatomical.
As shown below, XCP is attempting to find the anatomical in fmriprep/subj/session/anat (e.g.,
fmriprep/sub-BLMTBI104/ses-bline/anat/*MNI152NLin2009*
). However, the fMRIPrep structure has that file at subj/anat (e.g.,fmriprep/sub-BLMTBI104/anat
).I've test this with and without the id1 column in the cohort file and it makes no difference. The same error occurs.
Cohort file
Design File Paste your entire design (
.dsn
) file between the triple backticksError message Paste your error message between the backticks
Runtime Information Docker tag latest (Dockerhub release 6 days ago)
Additional context Docker call
fMIRPrep File Structure