Toolbox MEG-ROI-nets

Resting-state network analyses can be performed with the ROInets package in OSL.

See also: MEG-ROI-nets on Github

Starting with source-reconstructed data and a parcellation defining several regions of interest (ROIs), group level network matrices can be produced from the resting-state data. Multivariate corrections for spatial source leakage are performed as standard.

To begin, type

help osl_network_analysis

There is also a template file

edit ROInets.example
edit ROInets.example_multiSubj

The code package also includes several functions for:

• conversion from correlations to z-stats
• conversion between probabilities and z-stats
• performing multiple comparison corrections using the false discovery rate
• constructing single time-courses to represent entire regions of interest
• various methods of orthogonalising vectors, e.g. to remove the zero-lag correlations introduced by source reconstruction methods

To start exploring, type

help ROInets.Contents

To see a full description of the method, see “Colclough, G. L. et al. A symmetric, multivariate leakage for MEG connectomes NeuroImage (2015)”.