Table of contents

• 1. Gazelle
  • 1.1 Gazelle quick start
• 2. Coconi
  • 2.1 Load & Microtarget
  • 2.2 Network Models
    • 2.2.1 Force Directed graphs models and Directionality
    • 2.2.2 Nodal annotation in Graph Models
    • 2.2.3 Graph Thresholding
    • 2.2.4 Propagation models
  • 2.3 Nodal Centrality
    • 2.3.1 Nodal Centrality Ranking
    • 2.3.2 Average Temporal Centrality
    • 2.3.3 Nodal Temporal Centrality
  • 2.4 Coconi Export
    • 2.4.1 Save Figure

1. Gazelle

Gazelle is the app that will calculate synchronisation in the time-frequency domain across (a) a selected range of areas and (b) over selected annotations in the recording and (c) around a selected time window around those.

1.1 Event Loading

• Select file format ({.cnt, .edf+}, data files must include at least 10 annotations)
• Click Load EEG events to load your file from directory

A pop-up wizard will help you perform a, b and c above.

• In the middle column, select the type of event for analysis.
• In the right-hand column, you can further select events according to their elapsed time from recording onset.
• In the bottom, select the desired epoch window for analysis.

To construct a solid statistical basis for your analysis, You will then be required another two "baseline events". Ideally, baseline events are pre-annotated during EEG scoring so that they are (1) from background EEG activity (2) far from each other (ideally beginning and end of the recording).

• Click on Load EEG baseline events
• Select two events based on the 1, 2 criteria above
• Select all channels

1.2 Statistics

As a rule of thumb, a minimum of 1000 boostrap permutations is required for robust significance testing.

• Select the number of permutations for analysis
• Select level of significance

Hit calculate to get synchronisation maps. This step may take from minutes to hours, depending on the sampling rate of the records and the number of selected channels.

2. Coconi

After you have successfully calculated your connectivity maps, a NGmeta.mat file has been exported to your local directory. This file is consumed by Coconi and contains all the information for network meta-analysis.\

2.1 Load & Microtarget

• Add subject(s) to your analysis by clicking Add
• Start your analysis by clinkc Load

Microtargeting is all about looking for networks that are active in a specific frequency range (e.g. theta oscillations network) or time (e.g. pre-stimuli) range. In the microtargetting panel:

• Use the knob or value fields to target a specific frequency range.
• Use the knob or value fields to target a specific time range.

2.2 Network Models

There are two available options for network representation. In heatmap you can see analytically the connection strength between two areas coded in colour.

• Click on model tab.
• Select heatmap and the calculate to see connections and respective strength in colour density.

2.2.1 Force Directed graphs models and Directionality

Force Directed Model representations helps with understanding the positioning of each node within a network with respect to its relevant connections and centrality.

• Select force directed and calculate to see connections in lines, strength of connections in line width as well as positioning of nodes based on their inter-connetedness status.
• Turn on the directionality switch to see a directed graph representation of your networks

2.2.2 Nodal annotation in Graph Models

In the Force Directed Graph models, you can from a range of centrality measures to annotate nodes according to. This will help you identify where influential nodes sit within your network.

• In the Network Maps panel, select from the second from top dropdown menu the desired centrality measure.
• Click calculate.

2.2.3 Graph Thresholding

You can trim network connections by using a threshold value. In this way you can see how easily the network under investigation can collapse into smaller partitions/communities.

• In the network edges panel, use the slider to set a new threshold.
• Click calculate.

2.2.4 Propagation models

As an exploratory feature, you can model how activity is propagated through your network using the dendrogram map. 🚨This calculation is based on the inter-area phase difference and can be very unstable under circumstances - use with caution 🚨.

2.3 Nodal Centrality

You can use centrality measures to estimate the influence each of the nodes has upon your network.

2.3.1 Nodal Centrality Ranking

In the Rank tab:

• Select a centrality measure from the metric dropdown menu
• Hit calculate ⚠️ You may get an error message, if the centrality measures is not suopported by directed/non-directed graphs. Make sure to turn on/off the directed switch before you try again.

2.3.2 Average Temporal Centrality

You can do a dynamic analysis of your network, from a centrality point of view. For example, you can see how the overall centrality of the network may increase prior to the onset of an annotated event. In the Rank tab:

• Select a centrality measure of preference.
• Turn on the dynamic switch.
• Key-in a step value.
• Hit calculate.

2.3.3 Nodal Temporal Centrality

Instead of an overall centrality progression, you can also see how nodes may individually change their behaviour over time, with respect to their influence on the network. In the Rank tab:

• Select a centrality measure of preference.
• Turn on the dynamic switch.
• Key-in a step value.
• Turn the bottom knob from AVG to ALL
• Hit calculate.

2.4 Coconi Terminal

You can clear all, zoom ( select area to zoom in, double-click reset zoom), export a dashboard calculation to a figure or terminate the app.