BIDS Folder Strucutre

BIDS stands for brain imaging data structure. Since its introduction in 2015 it has become a standard on how to arrange imaging data in folders and subfolders, including naming conventions. Before BIDS, there was no consensus on how to structure imaging files. One can imagine that this lead to a lot of confusion. Furthermore, scripts that worked for one data set would not work for another, as the files were arranged differently. Having a standard solves this problem: Analysis scripts that are based on BIDS are now more or less universally applicable. This is why using the BIDS-format offers a number of advantages:

● Collaboration is easier. Different researchers can work on the same data set without any need for explanation on how it is arranged.

● You can use BIDS-Apps to analyze your data. BIDS-Apps are data processing pipelines that understand the BIDS format (more on that below).

● Data sharing is easier. Neuroimaging databases such as OpenNeuro.org require the data to be uploaded in the BIDS-format, so that they are understandable for everyone. For more regarding data sharing see the chapter on Open Data.

But how do we make our data BIDS-conform? Luckily, there are a number of converter tools that will do that conversion automatically (e.g. HeuDiConv for fMRI, plus a tutorial here). In the context of fMRI, you will generally need the original DICOM files for these to be able to work. Find the full list of converters from various formats on the BIDS website.

Now that you have converted your data, you might want to check if the conversion actually worked. For this, you can use a BIDS-App called BIDS-Validator. Additionally to checkingthe quality of the conversion, it tells you if there are any missing values in your data set. It can be run directly in a browser (Chrome or Firefox): To do this, visit this website and select the folder to your data set. Don’t worry—no data is uploaded. Alternatively you can run the BIDS-Validator in the command line, in Docker or in Python.

But BIDS-Apps can do a lot more than that, for example:

● Quality Control: MRIQC.

● Anonymization: BIDSonym. De-identifies fMRI images using one of several algorithms. They all basically crop the part of the image that contains the face, so that it can not be reconstructed and used to identify the person. De-identification is needed when you want to share your imaging data.

● Preprocessing: fmriprep. Yes, BIDS-Apps can even do the whole preprocessing for you. The philosophy of fmriprep is that of a glass box: the processing steps are automated and standardized, but you can see exactly what is going on because of quality reports and a thorough documentation. It uses the best tools that are currently available, by combining algorithms from different software packages.