search

PennLINC / xcpEngine

Official public repository for the XCP Engine. This tool is deprecated in favor of XCP-D and ASLPrep.
MIT License
66 stars 42 forks source link

NaN values for the regional ALFF and ReHo outputs #418

Closed llevitis closed 3 years ago

llevitis commented 3 years ago

Hi! Sorry for posting another question here; I just wanted to confirm that it's expected for some regions to have NaN values for ReHo and ALFF. It's my first time working with these metrics, and I was wondering if it would be acceptable to set those values to be 0 for subsequent analyses. Thank you!

llevitis commented 3 years ago

@a3sha2 I'd appreciate your guidance with this :)

a3sha2 commented 3 years ago

I am sorry for late response @llevitis Please let me know which of the regions and/or atlas pls. is it outside brain?

llevitis commented 3 years ago

Hi @a3sha2 here's an example using the Glasser parcellation where each row is a subject (I post-hoc merged the individual regional ReHo outputs myself)

image

a3sha2 commented 3 years ago

this is not expected. pls can you share the log file of one the subject with me pls

llevitis commented 3 years ago

@a3sha2 fc_levitise2_202012031421_sub-NDARINV8XENUGU9_ses-v01LOG-process.txt

I added the '.txt' extension so that I could upload it here. Please let me know if you were referring to a different type of log file?

llevitis commented 3 years ago

hi @a3sha2 - was that the right file or do you need another one? Please let me know if there's any other info I could share :) Thanks again for your help!

a3sha2 commented 3 years ago

Hi @llevitis sorry for late response. fmriprep functional output in T1w or MNI space are always in native space resolution. There is likely that your bold space resolution is very big. The main issues is that, xcp always does ROI extraction in bold input space which means that atlases (in MNI template space) are registered or resampled to bold input space. Then, Glasser atlas has many very small parcels and during registration to bold space, it is likely that some parcel disappeared due to big resolution of bold input space.
I will be glad if you can share one of your data

llevitis commented 3 years ago

@a3sha2 Oh interesting... I ran this using the MNI space output from fMRIPrep. Would you recommend I do this in T1w space instead? What's your email?

a3sha2 commented 3 years ago

Yes, if your output is in MNI space from fmriprep, the bold still retain resolution of (original)native space (not resolution of MNI) my email is azeez.adebimpe@outlook.com