Setup ICA Analysis
Before entering the parameters using Setup ICA, figure out how many
data-sets you are analyzing. Use batch script to run the analysis for large
data-sets.
Setup ICA Analysis step involves entering parameters for the group
ICA analysis. When you click Setup ICA Analysis button, a figure window
will open to select the analysis output directory. After the directory is
selected, parameter options figure (Figure 1) will be displayed. The parameters
in the figure will be explained step by step and for additional information
click on the link highlighted. Some of the user interface controls are shielded
from the user. These controls will be shown when you click
SetupICA-Defaults menu.
Figure 1: Initial Parameter Select Window
Main User Interface Controls:
- Enter name (Prefix) of Output Files
- All the output files created by EEGIFT will be saved using this prefix.
Enter 'Oddball'.
- Have You Selected The EEG Data Files?
- Select
- EEG data needs to be selected. When you click this button, you can
select the data using two methods.
- Method one can be used to select large data-sets. If your data is in
one folder and have a common file pattern, specify the filter pattern and
select the appropriate answer for the question "Are session folders inside
subject folders?". See screen shots
of the example data selection.
- Method two can be used to select data-sets that do not have data in one
folder or have a common file pattern. For large data-sets use batch script
for entering the data.
- After selecting the data files, the information is stored in a file
with suffix "Subject.mat" and the button will be changed to pop up window
with 'Yes' and 'No' as options. The explanation for 'Yes' is explained
below. 'No' is used if you want to select the data files again.
- Yes
- 'Yes' option will be shown if you have already selected the data files
based on the output prefix entered.
- The data reduction parameters are updated.
- If the data is entered for the first time, the numbers for the
principal components are set to 20.
- Number of IC
- Number of independent components extracted from the data.
- Do You Want to Autofill Data Reduction Values?
- Numbers for principal components are auto-filled. You can take a look at
these parameters by clicking on "Setup
ICA-Defaults" menu.
- Which algorithm do you want to use?
- Presently there are 10 algorithms in the toolbox that can be used for EEG
data like Infomax, FastICA, Erica, Simbec, Evd, Jade Opac, Amuse, SDD ICA,
Radical ICA , Combi, ICA-EBM and FBSS. A detailed explanation of algorithms like Infomax,
FastICA, Erica, Simbec, Evd and Jade Opac is given in the GIFT manual and in
related journal publications. By default, we employ Infomax since this
algorithm has been used most frequently in electrophysiology and neuroimaging
to our knowledge. The performance of the different algorithms is not
remarkably different when applied to aggregate EEG time domain data
- Note: ICA options window (Figure 2) will popup depending on the
algorithm selected. Presently, ICA options are available for Infomax, FastICA
and SDD ICA. ICA options window can be turned off by replacing variable
ICAOPTIONS_WINDOW_DISPLAY to 'off' (Defaults).
- Which Group ICA Analysis You Want To Use?
- Options are 'Regular' and 'ICASSO'. When you select 'ICASSO', ICA is
run several times and the best estimate for each component is used.
Figure 2: ICA Options Window
Figure 3: Completed Parameter Select Window

Figure 3 shows the completed parameters. Press Done button once the
answers for the parameters are entered. Parameter file for the analysis is
created in the working directory with the suffix '_ica_parameter_info.mat'
After the parameters are saved in a file, use
Run Analysis button to run the
analysis.