My colleague tries read an EEG signal, but in my opinion she is making a mistake. Her experiments fail. This device is using wet EEG electrodes (not reinforced, AgCl). In my opinion, the reliable measurement we can get only for bald places of a skull. What is your opinion?
Unusually thick hair can physically interfere with good coupling between the electrode and scalp when using gels and other electrolytes. But normally hair is not a problem for EEG. It is easier to place electrodes on bald patches but in my experience the impedance are often higher there.
Thank you for a quick answer. She tries make a simple experiment about idea of a hand movement, but her experiment has failed. I was thinking that a hairy place of a skull was a reason.
She recorded 100 samples with idea of movement of the left hand. Then, she recorded another 100 samples with idea of movement of the right hand. The samples were recorded for the two channels (C3 and C4). She averages samples and obtains:
A - The mean signal for C3 with movement of the left hand,
B - The mean signal for C3 with movement of the right hand,
C - The mean signal for C4 with movement of the left hand,
D - The mean signal for C4 with movement of the right hand,
Afterwards, she used FFT (Fast Fourier Transform) to obtain the Mu band (8-13Hz).
Finally, she computes power of the signal (formula is attached).
In Figure 1, she drafts Power of A and Power of B in the temporal domain.
In Figure 2, she drafts Power of C and Power of D in the temporal domain.
In her experiments in Figure 1 and 2 the differences are negligible, but for data from repositories everything is OK.
My conclusions were that the procedure of recorded data must be incorrect. I thought that hairy place of a skull can be the problem.
The last thing is an idea of movement of a hand. How can you imagine the movement of a hand. Maybe it is the real problem.
It could be a problem with imagination of the movement. How can you be sure that your subjects really imgined 100 times the movement? We have the same problems.
Off the top of my head, I can think of two main reasons that the readings differ from your expectations:
- There is difficulty in focusing one's imagination on one movement alone. What you're asking for from your subject is pretty much impossible to achieve in practice, and the best you can do is adjust to compensate for the unwanted readings.
- Repetition of the process can result in a programmed response by the brain. After enough repetition, it's not entirely impossible that imagining the movement of one arm triggers a response that emulates the movement of the other arm.
Thank you for the answer. This research is done by my colleague. Now I think, that I must ask on repository. Maybe there I will find the real reason of this situation. In the repository should be included the description of this research.
I don't think the problem is due to hair: paying attention during the preparation of your subjects, you could easily move hair under the electrodes.
As Elizaveta says you cannot monitor the real action accomplished by your subjects during the imagination task, after all it's not easy to do it (maybe impossible?).
I think you should check the analysis algorithm: you say that you (your colleague) averaged 100 epochs and then performed FFT to obtain the power of the mu rhythm. I would operate in the opposite way: firstly I would calculate the FFT and then average all the obtained spectra. It is not a good idea to calculate a power spectrum with FFT on a single averaged epoch (in this case a wavelet approach would be more suitable).
Moreover I would suggest to evaluate the event related desynchronization comparing, for each channel, the power of "action" epochs (movement imagery) with the power of "rest" epochs (i.e. 2-3 s before the action) and not comparing the same channel related to different movement sides.
Thank you very much for the answer. If this approach don't give a correct result, then the problem will be probably one of artifacts.
- Badges
- Science topic