The reference electrode is CPz. It is found that the signal amplitude of CP1/CP2/POz/Cz in some subjects is tiny in the preprocessing stage (as shown in the figure below). This may be due to the proximity of the electrodes to the reference electrode, or the electrode jelly causes unreasonable connections between these electrodes. How to deal with these electrodes? Keep them? Or remove them?
If they are retained, it may not provide much information, since the signal amplitude is extremely small, and it can be considered that the rank is reduced. If removed, they appear to carry certain information in terms of the spectrogram. I am very puzzled about how to deal with these electrodes reasonably?
Another problem: The waveforms of some channels of some data are almost the same, and they are almost the same from the frequency spectrum. This may be due to the problem of improper electrode contact. Should all these electrodes be removed?
Problem 1: I would remove the reference electrode from the dataset, and I would also keep the non-reference electrodes.
Problem 2: It seems that the signal is identical in the three electrodes. Are you sure that's not something you created artificially? Because even if the signal is overlapping across channels due to improper electrode-setting, there would some variability. My advices are : a) Use less gel, so it wont overlap between two channels, b) try spatial filtering (Laplacian for ex., matlab code here: https://github.com/vncntprvst/tools/blob/master/AnalyzingNeuralTimeSeriesData/code/laplacian_perrinX.m )
András Puszta Thank you very much for your answer. For question 1, I now prefer to keep these "flat" electrodes. For question 2, these data were not collected by me, so I can only guess the reason. The signals of these electrodes are almost the same, and using Laplacian may not solve the problem.
Eduardo Perez-Valero
Thank you for your answer. It is indeed necessary to compare the results .
As you said, the problem is either the proximity of the active electrodes to the reference, or the presence of salt bridges.
Even with the best possible technique in electrode appliance, the amplitude of the resulting channel would still be low.
In order to increase the amplitude of the signal, you should choose a reference outside of the isopotential map field. Ideally it should be far away from the head (like the neck or thorax...), but then you shoud be careful with the EKG "contamination".
Think about PSG montages. They are usually central electrodes referenced to the contralateral mastoid or ear (far away electrode). The reason for this is to maximize the amplitude of the signal.
Hope this helps!
Andre Schnoor Giorelli I can not agree with you more.Although we use any electrode as the reference, then we can re-reference. 'In order to increase the amplitude of the signal, we should choose a reference outside of the isopotential map field.'
It is annoying. The data has been collected, and whether these flat electrodes should be removed is a 'question'.
Do you know if the electrode impedances were measured before & after the data collection? if yes do you know these values?
Franklin Lue I know the electrode impedances before the data collection, and there are no abnormality in the electrode impedances of these 'flat' electrodes.
What are these electrode impedances? and were they measured before & AFTER the recording?
Franklin Lue most of them below 10 kΩ. they were only measured before the recording.
If you are unsure of the end electrode impedances, then it's quite possible that those electrodes came loose during the procedure....i have seen this on many occasions; i have also seen it being intermittent during procedures, when the electrode is not firmly in place.
Interpolate the signals of these channels from nearest neighbors or by similar techniques. E.g. https://mne.tools/stable/auto_tutorials/preprocessing/15_handling_bad_channels.html
and https://mne.tools/0.15/manual/channel_interpolation.html
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Electrodes