Author/Maintainer: Dianne Patterson Ph.D. dkp @
Date Created: 2018_12_07
Date Updated: 2019_09_14
Tags: BIDS, standards, containers
OS: UNIX (e.g., Mac or Linux)

Learn about this two-pronged project which describes (1) a standard file and directory naming structure for neuroimaging datasets, and (2) containerized apps that take assume this naming structure to great advantage. Below I provide useful links, including descriptions of software we’ve actually tried here at the U of A and related gotchas and tips.

Relevant Papers

Creating BIDS Datasets

Tools available for creating BIDS datasets focus on converting DICOMS to NIFTI and creating the accompanying json sidecar files for each NIFTI image. If you don’t have the original images off the scanner, there are not currently any good solutions for converting your data to BIDS. Part of the reason you should start with the DICOM (i.e., dcm or IMA images here at the U of A), is that there is considerable information in the DICOM headers that does not get transferred to the NIFTI headers. If you convert the DICOMs to BIDS, then this extra (deidentified) information gets saved in the json sidecar. The defacto tool for conversion is dcm2niix by Chris Rorden: download dcm2niix and read more about dcm2niix. dcm2niix produces the BIDS format json sidecar files, handles dcm and IMA files, and correctly saves slice timing for multiband data (which is not preserved in the NIFTI header).


Neither dcm2niix nor HeuDiConv will name your output files in a bids compliant way by default. You must look at the BIDS specification and figure out the naming. That said, I try to provide examples and guidance for UofA data on the HeuDiConv page.


The BIDS specification is under active development. In general, if your data is named in a compliant way today, it’ll still be okay tomorrow. However, tomorrow may bring additional options.