Flexible file formats.
One of the goals of The Virtual Brain is to allow users from different backgrounds to have quick access to their recorded neuroscience data and to openly share data with other specialized tools and applications.
From the abundance of common formats for neuroimaging data, The Virtual Brain supports the most popular ones. The TVB software also simplifies data exchange between TVB users with its own, documented data format (based on HDF5).
All file formats supported for data import are thoroughly documented.
Computational brain modeling tools such as The Virtual Brain require highly structured, self-consistent and multimodal data sets. Every year, large amounts of multimodal neuroimaging data are acquired all over the world. But extracting the necessary information, reducing the complexity sufficiently to create such data sets remains a challenge.
Considering the massive size of neuroimaging data, even for only a single patient, automated preprocessing pipelines for this analysis and transformation are needed.
Scientist from two TVB research facilities in Berlin (Petra Ritter and her team) and Marseille (Viktor Jirsa and his team) have developed pipeline solutions to produce data sets which can be imported into the TVB software:
The Empirical processing pipeline for TVB takes structural, functional and diffusion-weighted magnetic resonance imaging (MRI) and optionally electroencephalography (EEG) data. The code is publicly available on ted and diffusion MRI data to prepare surface, region mapping and connectivity data for TVB simulations. This code is also publicly available on GitHub.
The Surface and Connectivity Reconstruction Imaging Pipeline for TVB Simulations (SCRIPTS) uses T1-weighted and diffusion MRI data to prepare surface, region mapping and connectivity data for TVB simulations. This code is also publicly available on GitHub.
This suite of preprocessing modules was developed in the research project "How do parcellation size and short-range connectivity affect dynamics in large-scale brain network models?".
These solutions are under active developement and will be further refined until they reach the maturity of a ready-to-use companion tool to TVB.
Since working with brain activity data from real patients is obviously a highly confidential matter, The Virtual Brain lets you keep full control over the data you're working with.
All your data is only stored locally within your copy of the TVB software and never uploaded to any server of the TVB organization or other institutions.
When you are working on a project within TVB, all the data you generate with simulators and analyzers is stored internally in the HDF5 format.
You can export an entire project as a ZIP file. This ZIP file contains a folder structure which follows the data types you see in the TVB software. An embedded XML file serves as a guide for this folder structure. Each data type in itself is written into a HDF5 file.
It's also possible to export individual data types from TVB, e.g. just a connectivity, a time series or surface. These data types are, again, exported as single HDF5 files.
Exported projects can be imported again by other TVB users. The individual HDF5 files can be processed further by other neuroscience applications or by your own code. For popular tools and languages like MatLab, Octave, Python, Java or C++, there are libraries for accessing HDF5 files.
For processing HDF5 data in MatLab, there's an extensive API documentation available.
When you have configured a simulation within TVB, you can export the configuration setup as a JSON text file.
This is a real time-saver when you made a complex configuration and want other researchers to compare or reproduce your results.
Every visualizer within TVB offers a button to save its display in your project's image archive.
You can download these images as high-resolution PNG files. Some of TVB's visualizers are built with vector graphics. Archived images from these can be exported as SVG files that offer unlimited resolution.
By using exported TVB graphics in your research paper, instead of screenshots, you offer much higher quality to the reader.