How can I know the EEG signals I'm recording are correct? (Beginner Question)
Amir3022
Egypt
Hello everyone how are you doing?
We are a team of Electrical Engineering undergraduates at Cairo University, and we chose as our graduation project for our final year to make a BCI system using EEG signals recorded using an OpenBCI ultracortex IV headset with 8-Channels and Cyton board which were provided by the college.
We are intending to use SSVEP to stimulate the brain to generate a certain signal. Then use the recorded signal after processing the RAW data and applying filters as an input to a 1D-CNN model to classify the data.
This is for the most the approach we are trying to make here to make a simple BCI system that we can then use to make a virtual keyboard or even a Brain Controlled Wheelchair.
The problem now is we have very little experience regarding neuroscience and EEG in general. So we can't verify if the signal we are getting is even correct or not. From my knowledge that using SSVEP should produce a peak in the FFT of the EEG signals at the frequency at which the square on the screen is flashing with. But when we tried to replicate that scenario, we didn't get that result. and there was no significant peak anywhere on the FFT graph.
We also read that when closing your mind and going to a relaxation phase, the brain produces Alpha waves, which peak around 10 HZ, we also tried to replicate that but we didn't get the desired or expected result.
So I am asking if there is a way or a simple test we can do to make sure that readings are correct, and that we aren't just reading random noise?
One last test we tried was the blink and jaw clench test. Both created some sort of signal change in EEG recorded from Fp1 and Fp2, but no change on other channels.
Thank you for your time. Any help would be greatly appreciated.
Comments
Amir, hi.
Since this headset might have been used previously, check the condition of the silver chloride dry electrodes. You should not see any silver or white showing at the tips / edges that contact the scalp. If a color other than grey is showing, it means the layer of chloride has worn off the silver plate. Then the electrodes will not perform as well. You can also check the electrode impedances by pressing the ohm button (next to the channel number on the GUI). This should show impedance no higher than explained in this thread,
https://openbci.com/forum/index.php?p=/discussion/2517/expected-impedance-range-with-ultracortex-cyton-resolved
re: alpha test. Individuals vary in their production of alpha. So try different subjects. Be sure you are looking at the rear electrodes such as O1, O2 or P3, P4. Alpha is strongest in the rear.
re: SSVEP flash generation from LCD laptop monitor screen. This is not so easy, depending on how you are doing it. See some of these related threads:
https://www.google.com/search?as_q=ssvep&as_sitesearch=openbci.com
Generating flashes programmatically in real-time, can create 'jitter' in the frequency, because of OS, buffering, and other computer delays. Some people have had better luck creating a video file with the flashes, because video frame rate is more tightly controlled.
Further verification of your headset and Cyton can be done by following the tutorials on the docs site,
https://docs.openbci.com/docs/01GettingStarted/01-Boards/CytonGS
https://docs.openbci.com/docs/01GettingStarted/GettingStartedLanding
Regards, William
Thank you for your reply really appreciated.
I will read more into all the links and threads you have provided to me, and see if maybe I can do better next time I use the headset.
I am still searching for the SSVEP thing. I am not even sure I am doing it correctly. But I thought maybe that would be a quick way to test if the readings are correct, cause as mentioned, the headset and the kit are 2 years old and they were used by many teams before us, so it's not guaranteed it works properly.
I have already followed both setup guides for the Cyton board, as well as the guide to read EEG data and what to expect from them. And I got expected results. But still, whenever I try anything that should produce results in the FFT graph like the Alpha waves test or the SSVEP test, I don't get any results. So I was concerned maybe I was doing something wrong, or the kit is just not working.
As for the electrodes, all the electrodes still have the chloride coating intact. And we made sure to clean the forehead of the user with alcohol before wearing the headset to avoid skin oil having any effect on the impedance of the electrodes, so I guess the electrodes are working properly.
If you could provide me with a simple test I can replicate to make sure I am getting correct results regarding the FFT plot.
I will post again some screenshots of me trying the relaxation test again as soon as I get to the college campus again, so maybe you can help me figure out what I am doing wrong here.
Thank you for your help, and have a nice day.
If the headset has been used for multiple teams and years, then the tips of the electrodes could be worn. They are easily replaced. You can get extras of the short combs and cones on the FRI site.
https://www.fri-fl-shop.com/product-category/electrodes/eeg-electrodes/
The long combs are also listed there, and on the OpenBCI Shop. But the long combs also wear out faster.
https://shop.openbci.com/collections/frontpage/products/5-mm-spike-electrode-pack-of-30?variant=8120433606670
As mentioned previously, people differ in their closed eyes alpha generation. Please try other subjects than those who are not generating much.
I will try recording new data tomorrow from different subject to see if that will make a difference in the data we generate. As well as checking the dry electrode combs and making sure they aren't worn out. And maybe we will consider buying new ones just in case, cause we will definitely need some spare ones.
I will record the new data and post them here, and keep you updated about the whole thing.
Again thank you very much for your time.
These widgets will help to see if alpha power is increasing with eyes closed.
https://docs.openbci.com/docs/06Software/01-OpenBCISoftware/GUIWidgets#band-power
https://docs.openbci.com/docs/06Software/01-OpenBCISoftware/GUIWidgets#fft-plot
Another test you can do if you have access to cup electrodes and Ten20 paste, is compare the test results with 'wet' vs 'dry' electrodes. The signal amplitude is somewhat higher with wet based electrodes. Also impedance is lower with wet.
I have just added the Documentation for the Spectrogram Widget. This can be a very helpful widget when testing a variety of setups:
https://docs.openbci.com/docs/06Software/01-OpenBCISoftware/GUIWidgets#spectrogram-widget
You can see this in action here: https://openbci.com/community/gui-v5-release/
Take Care!
RW