time series after normalization / similar shape ?
I am currently using the Cyton board and WIFI shield while working on the MI_BCI project. After preprocessing and normalizing the data, I noticed that some channels exhibit similar trends. For example, in the figure below, all channels except for C4 and F8 show a similar shape. My question is: How can I interpret these channels to determine if the recording is incorrect or correct?
Comments
Kaveh, hi.
Can you clarify your OpenBCI products origin? Did you purchase these from OpenBCI in US, or are these 'clones' made in China? OpenBCI has not sold the Wifi Shield for quite a long time. That product is due to be refreshed or modified to correct some noise issue when used with Cyton. But timeline for that has yet to be determined. Some of the clone makers claim they applied the power supply mod mentioned below. But that does not really resolve the issue. There are many related threads here on the Forum,
https://openbci.com/forum/index.php?p=/discussion/2108/measured-shield-regulated-3-3v-supply-has-6v-dip-at-1-second-intervals-with-resonant-ringing
I also do not understand your 'normalization' operation. What are the actual microvolts you are seeing with these, in a normal times series graph? Your vertical scale gives no point of reference.
William
Hi Wjcroft,
Actually, I don't know because I recently joined the BCI group, and they have been working with this setup for almost two years. I can clarify it on Wednesday after I meet the company supervisors. Here is a picture of the Wi-Fi shield. Can you tell me if it is an original or a clone?
it is the sensors microvolts:
re: first photo
Yes this is an actual Wifi Shield, that has not been sold for a long long time. Again, the link I mentioned shows that Cyton does not work well with this shield. But Ganglion does work well. Your strange 'normalization' results may be showing the coherence of the noise pulses.
re: second image large microvolts.
Here you are seeing the DC offsets, which are normal in Cyton. The recommended way to remove the offset is either with a highpass filter at say .5 Hz, or a bandpass, say .5 Hz to 45 Hz. See this thread:
https://openbci.com/forum/index.php?p=/discussion/201/large-millivolt-data-values-fbeeg-full-band-eeg
Thanks for responding, I will do that and send you the result