Eye Blinks not detected

I ran Detect Eyeblink and found one huge deflection at 92 in VEOG not being marked as an eye-blink event.
I have marked it as eye-blink manually (pink). Do you advise me to go through all of the recordings manually and mark eye-blinks,cardiac or movement events? Is there any other less cumbersome way out ?

I performed SSP decomposition; and I guess SSP2 is a better component to remove;

However; After removing component 2; there seems to be an increase in amplitude in the highligheted channels.

Can someone help me - what has to be done here?

It looks like the first component captures slow fluctuations in your data. If they are of no interest, consider applying a high-pass filter (above 0.1Hz) first before running the eye blink event detection, and then re-run the SSP operation.

Hi team,
In few of the OMEGA subjects, Blinks from VEOG are not being detected. I then, filtered the VEOG signal with {high pass:0.3Hz ; Low pass:40Hz} and didn't find any difference in eye-blink detection event's counts or SSP components.
I wanted to check if during detection/calculation of SSP for blink already filters the VEOG signal. As I see the example of EOG[1.5-15]... in this image.

Do VEOG/ECG signals need to be filtered before detection of artifacts and SSP/ICA calculations?

In general, no need to preprocess the VEOG traces before running the Detect Eye Blinks process. How clean are the VEOG traces in these individuals? Can you see clear blink occurrences?

Hi Sylvain,
Apologies for the late response. Many of the VEOG/HEOG Traces aren't very clean.
Multiple categories are formed and some peaks are not marked as eye blink events

I am surprised that the blink events shown in the list do not seem to register to clear deflections in the HEOG and VEOG channels. In any case the EOG channels seem very noisy.

Precisely - It is the case with many subjects.
The above snapshots are of filtered EOG, MEG channels.
What would you suggest for artefact cleaning in this case ?

Instead of using VEOG and HEOG as reference channels for these artifacts, pick a MEG channel (prob ove frontal areas) that shows them the most and run the SSP event detection and attenuation processes.

1 Like