Hi,

I want to use this opportunity to grab your attention for letting you know about the new "Connectome File Format" (CFF) for multi-modal neuroimaging data we are developing.
The CFF is a container format that uses XML to add a layer of data and metadata description to individual data files. This encapsulations are called connectome objects. The datafiles are stored using existing standardized formats (e.g. Nifti etc.). Connectomes objects can be: CMetadata, CNetwork, CVolume, CSurface, CTrack, CScript, CData, CTimeseries, CImagestack. (See attached file)
For metadata description, we use the Dublin Core Metadata elements, and open metadata markup language odML (g-node.org/projects/odml). odML provides flexible means to annotate
metadata grouped in sections, that contain properties with name, value, unit, uncertainity, type information.

We provide the Python library "cfflib" for reading/writing of such connectome file, providing an object model based on CFF to work with.
It exposes a common interface to all objects. It heavily depends on Nibabel for basic IO:
https://github.com/LTS5/cfflib

Example datasets:
https://github.com/LTS5/cffdata/tree/homo_sapiens_02/homo_sapiens_02

The CFF XML Schema:
https://github.com/LTS5/cfflib/tree/master/cfflib/schema

In allusion to the "Extending the Nifti Format" thread, we are planning to use the CFF XML file in conjunction with HDF5, i.e. integrating the XML file as HDF5 metadata and reference HDF5 paths as data sources instead of data files.

Any comments are welcome!

Stephan