To provide the tools necessary for reproducible analyses. The analyses must also generate queryable, machine-readable data and results using interoperable and standardized data models. We aim to:
Neuroimaging data have become richer, with larger cohorts of participants, and a greater variety of acquisitions. A large number of useful analysis methods are now available, and many pipeline tools make such analysis efficient.
Currently we are focused on the following projects. You can help us by trying them, contributing to them, or by sharing your ideas on how to improve them.
Describe experiment setup and data, analysis details, and results in a structured manner. This is a collaboration with the Neuroimaging Data Sharing Working group, to develop and maintain data models and software to support rich and structured description of research workflows.
These tools will include publishing, visualizing, and validating data described using NIDM. It will enable interoperability with existing efforts (e.g., BIDS, DataLad, Nipype). Example: Keator et al., 2017
ReproNim is developing BrainVerse to help researchers manage, track and share information in a comprehensive format. Brainverse is an open-source, cross-platform desktop application to enable researchers add to reproducible practices into every project.
Create a comprehensive framework for robustness testing. This will enable continuous evaluation of algorithms under different operating system environments. Neuroimaging relies on numerical computations and results can depend on:
Testing analysis in various computational environments allows to:
ReproNim's Testkraut implementation will allow researchers to continuously evaluate the performance of algorithms developed in the community and to choose the most appropriate tools for a given analysis task.