About ASCT

ASCT stands for Atlantis Source Connectivity Toolbox.

The Atlantis Source Connectivity Toolbox (ASCT) is an open-source software designed for a complete analysis of directional connectivity on EEG and MEG datasets. It is provided with the GNU General Public License (GPL).

The ASCT requires Matlab environment and is dependent on some other code, including FieldTrip , EEGlab , fastICA , SimBio , HCP Workbench , AAR , ROInetworks , and Multar.

During the development of the software, attempts have been made to test and determine the impact of parameters on the final results. ASCT comes with a set of recommendations according to the analysis setup, to provide replicable and reliable results.

Theory behind the ASCT pipeline

The detailed theoretical introduction to the analysis performed in the ASCT is described in the manuscript being under review.

The analytic pipeline is presented in the diagram below. After data import, preprocessing steps are performed. This includes filtering, trigger recoding (optional), segmentation of data, bad channel AR (artifact rejection), and trial-based AR. Next, ICA (independent component analysis) decomposition is carried out for the first time. It is intended to enable another stage of AR and remove artifacts that are better seen on the component than on the sensor level. After cleaning, the signal is again subject to ICA2 decomposition with dimensionality deflation enabled, which is performed multiple times, and later on, the best ICA realization is chosen for further analysis. ICs (independent components) are then classified into those of brain and non-brain origin. The latter are considered artifacts and ignored while the brain components are localized using the MNE method. Brain signals for the ROIs (regions of interest) are then reconstructed based on all brain ICs, taking into account their localization and timecourse. Localization can use individual head and source models based on MRI T1 scans or SPM templates if scans are unavailable. 3D reconstructed signals are subject to PCA (principal component analysis), where the first component is retained, representing orientation-independent dominant ROI activity. Next, the leakage correction is performed to remove zero-lag correlations from ROI signals that may result from the spatial blurring of localized sources. Finally, the directed connectivity is estimated using the Directed Transfer Function and the results can be visualized and exported.

Cite

You are free to use the software. If you find it useful, please cite it as:

Development Team

The toolbox has been developed at the Institute of Psychology, Jagiellonian University, Krakow, PL. Some parts of the code have been written in the Institute for Advanced Medical Technologies, Chieti, IT.

NextInstallation

Last updated 1 month ago