BIDS file naming specifications

bids

#1

HI, I’m attempting to validate our BIDs conversion for a set of macaque functional imaging data. Unfortunately, we’re getting an invalid file name error on the BIDS validator. Based on the output, I have two questions about the specifications. 1) do you have to have the word “task-” in your file name? 2) Is BOLD the only valid functional data. Number two is important as we are technically not collecting BOLD we’re collecting MION signals. Thanks


#2
  1. task is a required keyword for _bold, _events, _sbref, _physio, _stim and MEG related files. See Appendix IV for a breakdown of which keywords are optional and which are mandatory for different files. BTW please mind that “rest” is considered a type of a task.

  2. Current spec does not cover MION imaging, but it would be a great extension. Would you like to work on this topic? Here’s an outline on how to start https://docs.google.com/document/d/1pWmEEY-1-WuwBPNy5tDAxVJYQ9Een4hZJM06tQZg8X4/edit. Please also mind that when including data not yet covered by BIDS you can use the /.bidsignore file to instruct the validator to ignore selected files. https://github.com/INCF/bids-validator#bidsignore


#3

Thanks Chris,

I’ll definitely look into the extension, I would be happy to help with that. In the mean time, would you suggest us treating our MION data as BOLD? I.e “mislabeling” it as _bold, so that things like fMRIprep recognize it as functional data?


#4

I don’t know much about MION imaging to give an informed decision. Is it an EPI sequence with contrast injection? How does tissue contrast differ from normal EPI?

As for FMRIPREP please mind that it does not yet work with non-human data.


#5

MION is a iron contrast agent often used in non-human imaging. It provides a signal to noise boost, and isolates the blood volume component of the BOLD signal. It is a essentially a standard EPI sequence with a faster TE. The biggest difference is that at a processing level, a MION signal is an inversion of the BOLD signal. A drop in signal equates to more blood (iron) in the voxel.


#6

Thanks for the clarification. In this case I would recommend proposing a new file naming scheme.


#7

I think that makes the most sense in the long run as well. I’m looking over the Contributor Guide now.


#8

Dear Brian,

Is there any update on this extension? I was wondering how limited one can be in the use of the current BIDS Apps with non-human + MION data. Is there any of the Apps that you’ve found useful? Thanks!


#9

HI Sajjad,

I haven’t had any chance to work on the extension unfortunately. Currently we’re only using the niix2bids functions with a little bit of tweaking to deal with NHP orientations and a few other idiosyncrasies in the data. Thought we’re happy with that, and the data are now BIDs compliant.

One of my goals it to eventually try to run some of the MRIPrep applications but haven’t gotten there yet. At first pass it looks like the apps will need some tweaking to run non-human data, though I will be honest and I didn’t spend a long time on them as of this time.

I hope this helps.

Brian


#10

Thank you very much for your response, Brian. I’d need your help later with making my macaque data BIDS compliant; mine was also acquired at NIH!

I see. One app that I know would work with non-human data and might be useful for you too is PyMVPA BIDS which I just recently released. I’m making some final changes and it will be ready for use soon. I’ll keep you posted on how running it on my data goes.

Best,
Sajjad