Download REGICA plugin

 

File: REGICA plugin for EEGLAB

Uploaded:

22.02.11

Modified:

22.02.11

File Size:

43 KB

Downloads:

78

Version

1.0

If you use this plugin please cite:

Manousos A. Klados, Christos Papadelis, Christoph Braun, Panagiotis D. Bamidis, REG-ICA: A hybrid methodology combining Blind Source Separation and regression techniques for the rejection of ocular artifacts, Biomedical Signal Processing and Control, In Press, Corrected Proof, Available online 16 March 2011, ISSN 1746-8094, DOI: 10.1016/j.bspc.2011.02.001.

Download

 

 

 

License and disclaimer

Rights, ownership, and copyright information related to each MATLAB file included in the REGICA plugin are individually stated in each file. The current release is relatively stable and many bugs have been corrected. However, it is intented to be used for research and testing purposes only. No claims are made as to the validity of the methods or the correctedness of the plugin code and documentation. Bugs and suggestions can be reported to This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

 

Installation

Installation instructions:

1) Install EEGLAB toolbox for Matlab (http://www.sccn.ucsd.edu/eeglab/).

2) In the plugins folder of the EEGLAB installation folder create a folder called REGICA and place all the files of the REGICAv1.0 plug-in inside that folder.

3) Add the REGICA folder that you created in the previous step to the Matlab path

4) Run EEGLAB. Under the Tools menu there should be a sub-menu named  "REGICA" which contains the REGICA functions.

 

REGICA

BSS

REGICA consists by three parts. The first one is the selection of the BSS method which will decompose the EEG signals into statistical independent components (ICs).REGICA recognizes all the installed BSS methods in the EEGLAB (Fig. 1)

Fig.1 - BSS Methods

Identification Criteria

The second part deals with the selection of the artifactual ICs. In the REGICA(v1.0) there are two different ways to recognize the artifactual components. The first one is based on the fractal dimension which was firstly described in [ref],(Fig.2) while the second one computes the correlation coefficinets between the artifactual components with the EOG signals and then it filters only the components which their correlation coefficients are greater than a defined threshold (Fig.3).

Fig.2 - Identification criteria. (Fractal Dimension)

Fig.3 Identification criterion based on correlation and the correlation threshold.

 

Regression Algorithms

Five different regression algorithms (Fig 4.) for EOG removal using one or more EOG regression channels are used for the decontamination of the artifactual ICs. The algorithms are:   

• Least Mean Squares (LMS)
• Conventional Recursive Least Squares (CRLS)
• Stable Recursive Least Squares (SRLS)
• Algorithms based on thenorm

            (follow the links to learn more about the algorithms and their parametrization)

The aforementioned algorithms are part of the Automatic Artifact Rejection toolbox for EEGLAB.

Fig.4 - The listbox depicts the option for the user to select the regression algorithm

 

After the selection of the regression algorithm the user has to importi the indices of the EOG signals in the corresponding textbox. (Fig. 5) It should be mentioned that the EOG signals should be part of the EEG.data matrix. For example if the EEG.data is an mXN matrix (m:sensors N:sample points) then the (m-k)XN should be the EEG signals while the kXN should be the EOG signals respectively.

 

Fig.5 - The option of the selection of EOG signals is enable for all regression algorithms.

 

Regarding now the parametrization of each regression algorithm, the needed field are enabled, while the rest are all disabled. For example in the following figure:

Fig 6 - LMS parametrization

Only the M and mu parameters are enabled (green box) while the rest parameters which stand for another algorithm are all disabled (red box). In Fig. 7 you can notice that for the SRLS algorithm different parameters are enable or disabled.

Fig 7 - SRLS parametrization

 

Correction examples

 

Real EEG data

 

 

Artificially Contaminated EEG Data described in [2]

Test Dataset available for download

File: Artificially Contaminated EEG Data

Uploaded:

22.02.11

Modified:

22.02.11

File Size:

51 MB

Downloads:

31

Version

1.0

The dataset used in the [2] for the evaluation of the REGICA methodology is availiable for download. This dataset consist by 54 free of ocular artifacts EEGs (Pure_Data) namely sim{i}_resampled (where {i} is a number from 1 to 54), 54 VEOG (VEOG.mat) namely veog_{i}, 54 HEOG (HEOG.mat) namely heog_{i} and 54 artificially contaminated EEGs (Contaminated_Data.mat) which is the artificially contaminated data. More information for the contamination procedure you can find in [2].

If you use this dataset please cite:

Manousos A. Klados, Christos Papadelis, Christoph Braun, Panagiotis D. Bamidis, REG-ICA: A hybrid methodology combining Blind Source Separation and regression techniques for the rejection of ocular artifacts, Biomedical Signal Processing and Control, In Press, Corrected Proof, Available online 16 March 2011, ISSN 1746-8094, DOI: 10.1016/j.bspc.2011.02.001.

Download

(Needs free registration)

 

Studies used REGICA 

• E. G. Peranonti, M. A. Klados, C. L. Papadelis, D. G. Kontotasiou, C. Kourtidou-Papadeli and P. D. Bamidis, "Can the EEG Indicate the FiO2 Flow of a Mechanical Ventilator in ICU Patients with Respiratory Failure?" , PD Bamidis and N. Pallikarakis (Eds.): MEDICON 2010, IFMBE Proceedings 29, pp. 827-830, 2010

 

(If your study is not listed here, you are kindly pleased to send me an email with the citations of your work.)

 

 

Acknowledgments

 

This software package includes parts from the EEGLAB toolbox, as well as from the Automatic Artifact Removal v 1.3 (AAR1.3) plugin for EEGLAB developed by German Gomez-Herrero.

The artificially contaminated dataset were collected by the members of the Group of Applied Neuroscience. 

We would also like to thank Christos Frantzidis and Christos Moridis for their helpful contribution during the testing procedure of the plugin's beta version.