Software

Mutual Connectivity Analysis with Local Models

This software was developed using MATLAB* 2016a and contains all the functions required to perform mutual connectivity analysis with local models (MCA-LM) [1]. This method can be used to extract directed influence flow between every pair of time-series in a system. This toolbox demonstrates the applicability of MCA-LM on realistic fMRI simulations generated using the NetSim software [2]. Performance is evaluated by comparing the recovered networks with the true network structure (adjacency matrix) of the simulations. Effects of reducing repetition time (TR) are demonstrated in the demo.

References

[1] DSouza, Adora M., Anas Z. Abidin, Udaysankar Chockanathan, Giovanni Schifitto, and Axel Wismüller. "Mutual connectivity analysis of resting-state functional MRI data with local models." NeuroImage 178 (2018): 210-223.

[2] Smith, Stephen M., Karla L. Miller, Gholamreza Salimi-Khorshidi, Matthew Webster, Christian F. Beckmann, Thomas E. Nichols, Joseph D. Ramsey, and Mark W. Woolrich. "Network modelling methods for FMRI." Neuroimage 54, no. 2 (2011): 875-891.

Download

Download MCA_LM_toolbox

Large-Scale Granger Causality Toolbox

This software was developed using MATLAB* 2014a and contains all the functions required to perform large-scale Granger causality (lsGC) Analysis [1,2]. This method can be used to extract directed influence flow between every pair of time-series in a system. It is an extension to multivariate Granger Causality analysis for very large systems, such as functional MRI data sets. The principle of Granger causality is based on the concept of cross-predictability where the improvement in prediction quality of a time-series in the presence of another time-series is evaluated and quantified, revealing influence direction between the two series.

The flow of influence between individual time-series is obtained using predictive models such as vector auto-regressive (VAR) modelling. For systems with very large number of time-series N as compared to number of time points T (N>>T), calculating a multivariate Granger causality index is not possible as the parameter estimation in multivariate predictive modelling scheme is limited by T resulting in an underdetermined problem. To counter this problem, we introduce a linear dimension reduction step prior to performing VAR. Predictions are performed in the low-dimensional space which are transformed back to the original space using the inverse of the transformation function. This permits us to compare the prediction with the true time-series in the original space, resulting in an lsGC index for each pair of time-series in the system.

References

[1] DSouza, Adora M., Anas Z. Abidin, Lutz Leistritz, and Axel Wismueller. "Exploring connectivity with large-scale Granger causality on resting-state functional MRI." Journal of neuroscience methods 287 (2017): 68-79.

[2] D'Souza, Adora M., Anas Zainul Abidin, Lutz Leistritz, and Axel Wismüller. "Large-scale Granger causality analysis on resting-state functional MRI." In Medical Imaging 2016: Biomedical Applications in Molecular, Structural, and Functional Imaging, vol. 9788, p. 97880L. International Society for Optics and Photonics, 2016.

Download

Download lsGC_Toolbox.zip

Copyright Wismüller Computational Radiology Laboratory, November 2015 University of Rochester,  Adora D’Souza (adora.dsouza@rochester.edu)

*MATLAB is a registered trademark of Mathworks, Inc.