I do the data analysis in EEGLAB and want to do ICA to remove the artifact. As the EEG data is recorded in .bdf, so I have two questions:
1. There are no reference in the recorded data and when I import EEG data into EEGLAB, I had to chose one channel as the ref, this may cause the rank-deficient;
2. I had to deal with the bad channels before I do the ICA, but I find if I do the interpolation before ICA, the rank of the data also reduced (interpolate one channel make the rank-1).
As rank deficient will make mistake for ICA result, so does someone konw how to deal with these two problems? Thanks.
Hi wenwen,
1, I think you can use the average reference after loading the data into EEGLAB.
2. After you remove the bad channels, you can only decompose your data into components as many as good channels. After that, you can bring your bad channels back by interpolation.
Ping
Hi Wenwen,
Though my contribution is not related to EEGLAB , I suggest that you follow these valuable references in order to address ICA-based EEG data denoising efficiently:
- Albera et al., "ICA-based EEG denoising: a comparative analysis of fifteen methods", 2012 - https://hal.archives-ouvertes.fr/hal-00740304/document
- Romo Vasquez et al., "Blind source separation, wavelet denoising and discriminant analysis for EEG artefacts and noise cancelling", 2014 - https://www.researchgate.net/profile/Radu_Ranta/publication/47760548_Elimination_d'artefacts_et_du_bruit_d'EEG_epileptiques_debruitage_et_separation_de_sources/links/5745759308ae9ace8421b842.pdf
- Castellanos et al., "Recovering EEG brain signals: Artifact suppression with wavelet enhanced independent component analysis", 2006 - https://www.researchgate.net/profile/Valeri_Makarov/publication/6956279_Recovering_EEG_brain_signals_Artifact_suppression_with_wavelet_enhanced_independent_component_analysis/links/00b7d518942d5a7e4c000000.pdf
- Rivet et al., "Wavelet de-noising for blind source separation in noisy mixtures", 2004 - https://www.researchgate.net/profile/Christian_Jutten2/publication/220848392_Wavelet_De-noising_for_Blind_Source_Separation_in_Noisy_Mixtures/links/0c9605189548c4c5be000000.pdf
To denoise efficiently, ICA is commonly used in association with Wavelets. Refer to the above references. Moreover, guidelines are given for denoising without altering the original mixture model.
Regards
Article Recovering EEG brain signals: Artifact suppression with wave...
Conference Paper Elimination d'artefacts et du bruit d'EEG épileptiques : déb...
Conference Paper Wavelet De-noising for Blind Source Separation in Noisy Mixtures
@ping yang, thanks a lots.
So I can do the average reference by excluding the bad channels, right ?
And for the 64 channels' data if I do the average ref, the rank will be 63 and I will lost the data of this ref-channel, do you have good idea if I need 64 channels data?
Thanks a lots for your reference and information@Stéphane Breton
I will read these materials carefully and if necessary, discussion with you.
meric
Hi wenwen
I am confused about no reference in the recorded data, is it means you have 65 channels when recording the data and only can get 64 channels without the ref channel when exporting the data out? If so, it’s ok to do the following analysis I think (filter, bad trials/channels remove, ICA artifact remove, channel interpolation, average reference and baseline-correction). It doesn’t happen to me that EEGLAB require me to select one channel as a ref channel when I loading the raw data, I don’t know how to deal with that correctly.
Ping
@Ping, Thanks a lots for your information.
As my data is recorded in Biosemi system, and the reference is the CMS and DRL, this is just the common reference and no data channel recorded for the reference, and if I want to import data in EEGLAB, I have to choose one channel as ref and then do the re-reference. I am not sure do you use the Biosemi system before? Now I just choose one channel as ref and I have to lost 1 channel (I have 64 channel raw data and after re-average, I get 63 channel data, and I did ICA for the channels which exclude the bad channels, after that do the interpolation )
my another question is: if I do the ERP, my epoch has overlap, do you think it is better to do epoch before ICA or after?
Thanks for your patient and information.
Wenwen
Wenwen Chang in this case no need to select the reference in eeglab as you have already use CMS and DRL during the recording. But in eeglab you can select rereference the channel even after the recording.
Hi, Wenwen. Another possibility is run PREP-pipeline which is a plug-in of EEGLAB to robust-reference your data. You can specify the rank of ICA to avoid the deficiency problem. Hope that can help you.
- Badges
- Science topic
- Similar topics
- Correction