[Eeglablist] Eye blink counting

[Arnaud Delorme]

Hi Jan,

there is an EEGLAB plugin to detect eye blinks (CORRMAP distributed with EEGLAB). There is also a more powerful solution called EyeCatch implemented by Nima Bigdely <nima at sccn.ucsd.edu> (email him as I do not think his plugin is available on the internet). It will be able to detect your blink components in each subject.
Then the data is in the array

EEG.icaact(x,:)

where x is the component containing blinks.
As far as setting a threshold, I would personally implement a threshold that is unique to each subject (since the amplitude of the blinks vary from one subject to the other). You could make the threshold depend on the variance of the component activity for example.
Best,

Arno

On 19 Nov 2012, at 14:38, Tom Campbell wrote:

>  
> hi Jan,
>  
> As a start, I'd calculate a bipolar VEOG channel by subtracting SO1 and IO1 data (or nose and Fpz; or SO2 and SO1) that you can find by selecting the appropriate channels from the structure continuous EEG.data after filtering. wthresh (for thresholding) and SplitVec (downloadable from MATLAB central, for parsing output thresholds into groups) are handy functions as are size and length (for counting), together with figure and plot (for plotting VEOG and thresholds to assessing how well your algorithm detects blinks). VEOG amplitude threshold and the criterion minimum number of consecutive samples to reach threshold is likely best hand-tuned (individually) to prevent false alarms and misses, requiring manual inspection of the adequacy of the thresholded values, though when you are happy with this, counting can be automated. Blinks are biphasic, which is something to think about in your thresholding and parsing. You may decide to include other safeguards that the identified VEOG is not due to upward eye movements (not easy) or horizontal eye movements as assessed by HEOG (LO1 and LO2 difference; e.g. threshold only blink samples where the SO1-IO1 difference is greater than the LO1-LO2 difference) but rather due to blinks.
>  
> There is also a blink detection function in the ERPLAB plugin:http://erpinfo.org/erplab/erplab-documentation/documentation-archive-for-previous-versions/v1.x-documentation/erplab-manual/Artifact_Detection.htm
>  
> T. 
> 
>  
> 
> From: janinthesky at gmail.com
> Date: Fri, 16 Nov 2012 17:24:13 +0100
> To: eeglablist at sccn.ucsd.edu
> Subject: [Eeglablist] Eye blink counting
> 
> Dear EEGlab list, 
> I am trying to count the number of eyeblinks in my data. The eyeblink component after running ICA tends to occur at a different number in each subject's data. I would appreciate it if anyone could tell me 
> a) firstly how to isolate one component's data and view it as a matrix of time points and microvolt values so I can threshhold out the eyeblinks and
> b) if there is a way to automate the process of counting over all subjects (because the component number may be different in each) or if I will have to do this manually?
> 
> Thank you very much for your help, 
> Jan
> 
> 
> _______________________________________________ Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20121126/0e788634/attachment.html>