The derivatives folder will contain multiple different 'preproc' folders, holding preprocessed data. Each holds cleaned, preprocessed data for a given study or analysis. That is, this data that is above the level of raw or minimally-processed data, but is not the output of higher-level analysis.

This allows preprocessing to be readily accessed across multiple different higher-level analyses (i.e., redundancy is reduced). You will therefore have multiple preproc folders within derivatives.

As noted, we store minimally-peprocessed data in 'preproc_core,' including (for example) skull stripped structural images, origin-corrected images, and transformation matrices.

We then have preproc folders for each study–

  • /preproc_core
  • /preproc_study1
  • /preproc_study2

And so forth. The results of the analyses completed on data in this folder will be written to a derivatives/analysis... folder.


This folder contains a readme file, followed by a folder for each subject who has received this form of preprocessing. Each subject folder may contain a folder holding each session of data acquired.

Item Description
README Plain text file holding whatever information your team elects.
sub-y0001 Folder containing preprocessing for subject y-0001 This should be capitalized
sub-y0002 Folder containing preprocessing for subject y-0002

Note: Full details available at BIDS website; the full specification is definitive.

Subject level file names within preproc directories.

The subject/session folders here will contain both /anat and /func subdirectories for each individual; e.g. –

  • derivatives / preproc_[study] / subj-[id] / ses-[id] / func /

Note that in deriving names, the base is always the name of the sourcefile, with additional variant information appended to the file name. When deriving switches, consult the common derivatives [proposal][1]. Listings of the conventions and our standard file names are here –