By Yumie Ono and Adam Eggebrecht
Did you know that almost 2,000 fNIRS articles were published in the year 2018? In fact, the number of published fNIRS research has been rapidly growing in these two decades.
This means that fNIRS is being accepted as a valid method of brain functional imaging, but at the same time the fNIRS society requires to provide more opportunity for newcomers to learn from ‘typical’ datasets and discuss their quality of data acquisition due to its susceptible nature for systemic effect and light absorption by anatomical constraints. It is also important for already established fNIRS research groups to share their datasets to facilitate multi-centered clinical and academic collaboration.
Researchers in fNIRS community have been working since 2012 to establish a shared data format for fNIRS experimental data. Like DICOM format in magnetic resonance imaging, the shared near infrared (SNIRF) file format provides numeric fNIRS data with time stamps and a so-called ‘measurement list’ for detailed measurement information. Although the majority of fNIRS data types are continuous wave NIRS, SNIRF format also covers other types of NIR measurements, such as frequency-/time- domain, diffuse correlation, fluorescence, and bioluminescence imaging. Other important information such as trigger timing, channel/optode positions, source-detector separation, wavelengths of NIR light, source power, and detector gain can be included in the list. You can find more detailed file format specification and a list of contributors in GitHub snirf page (https://github.com/fangq/snirf/blob/master/snirf_specification.md ).
The SNIRF file has .snirf extension and is provided with HDF5 format (Hierarchical Data Format 5: https://www.hdfgroup.org/ ), one of the scientific data format designed to store and organize big data. The HDF5 format is accessible and editable by major software platforms of C, C++, Java, Python, MATLAB, etc, so would be compatible with existing fNIRS analysis software.
There are many NIRS manufacturers providing their own data format, which sometimes makes end-users difficult to use specific data analysis software and share data with collaborators. The SNIRF format has great potential to resolve these difficulties and expand the opportunity for research collaboration and systematic big-data project. Further collaboration of society members is required to provide example dataset and data conversion add-ons for existing analysis software.
Society member who is interested in contributing to further development of SNIRF format can contact to David Boas.