I'm not sure I understand your English. What is a "PPT Plot"??
FFT of EEG does not really tell you very much, although it can show which frequencies are predominant.
All I need 2-2 hours recordings, the range of uV form 20 to 30,000 on the scale,
I already told you that normal range of EEG is up to say 80 uV. 30,000 uV is pure noise. Or perhaps you are confusing the DC offset. Please see previous link.
I don't mean to be rude, but does your supervisor have any previous exposure to EEG or digital signal processing? What are you trying to achieve by collecting the FFT at certain moments? If you are seeing large spikes in uV amplitude in your time series, that usually corresponds to noise caused by movement, muscle artifact (EMG), headset motion, etc. etc.
GREAT. Thanks for finally explaining what you are trying to achieve (!)
The FFT does not indicate 'emotional' content. There is an entire field of neuroscience devoted to this. Try some of these search links and do some reading.
He was asking questions can we export time data (say 8 hours) in Excel. To me, it seems possible because the software does not put any limitations.
and other question was can we picl event in 8 hour recording and get FFT of event. To me, why now if it has been recorded. Maybe not in the scv, but MATLAB it should be in case it has been recorded. and FFT goes every 20 seconds.
There will be a qualified people involved for setting and analyzing experiments. and for my sake, I read everything possible because it makes me easier to prepare everything for the experiments. to me, range of 20-30,000 did not make sense and I kept calling everybody feeling my ignorance. So if I understand I could explain why it is possible or not possible. It drives me crazy because I just do not understand what should be measured. - https://fda.gov/media/74662/download
To reiterate, the FFT is not going to help you in any way whatsoever, to determine emotional valence or emotional state. By itself. Emotion recognition requires sophisticated DSP, such as is outlined in the research paper links previously given. This type of EEG research is still in early stages and DSP required is substantially sophisticated and complex.
PLEASE have your company and yourself do some reading on EEG and emotion. See previous links. You cannot do this with FFT alone. So your entire outline of what you want to do at this point, would be extremely frustrating for all involved. Putting it bluntly.
See, my boss is a scientist. I just got a job, and need to follow directions. So I can learn programming, or anything he needs from what I understand, and I am grateful for opportunity to learn. To me, those just requirements and maybe later I will understand why.
PLEASE have your company and yourself do some reading on EEG and emotion. See previous links. You cannot do this with FFT alone. So your entire outline of what you want to do at this point, would be extremely frustrating for all involved. Putting it bluntly.
@wjcroft said:
PLEASE have your company and yourself do some reading on EEG and emotion. See previous links. You cannot do this with FFT alone. So your entire outline of what you want to do at this point, would be extremely frustrating for all involved. Putting it bluntly.
Thank you! I understand. - if there are going to be scientists involved, they can be lay audience in terms of usage of the software. So it should be as simple as possible. And it seems to be my responsibility for now and then to fill in where it is needed.
It is my job to do all the readings and explain all things regarding the software. It was my idea to purchase it because I thought it was the best on the market for the purpose.
I cannot help you any further, unless you have specific OpenBCI questions. If you are interested in EEG and emotion, you need to do some background reading on how that is currently being done. FFT of noise bursts will not help you.
Thank you! I know. Maybe they want to measure the noise of aircrafts. I promise you I am doing all background reading and will do more. scale did not make sense to me. and I need to speak up and start asking questions why. So far I am focusing on what shall be done. Take care! I appreciate your help and patience.
The three types of potentials that can be measured are EEG, ECG, and EMG. The EEG only gets up to say 100 uV at the most (this is unusually high). Normally ranges from a few uV to below 80 uV. The EMG and ECG can get into millivolt range, (thousands of microvolts). 30,000 uV is 30 mV (millivolts).
With your special purpose emotional detection / recording, you will be using your own signal processing. Initial stage of that will filter out EMG and eye blink artifacts. Note that the raw EEG recording contains noise + valid EEG. There is usually no need to plot the EMG noise signal.
"FFT of EEG does not really tell you very much, although it can show which frequencies are predominant."
We need to have a window with the time, so when we use PlayBack widget, we can select a time for FFT event to see what happened during that time of the event.
I have not figure out yet alternative - which program to use to pick the event and get FFT of the event.
Please, give me suggestions.
It seems more like a programmer can add a time drop down or just where a user can add the exact minutes/seconds. Any suggestions? Thank you!
Lisa, hi. I changed the title of this thread to: "collecting emotions based on specific EEG events", for more clarity.
The easiest approach for capturing the EEG + FFT around the time period of a specific event, in general will NOT be to pursue making changes in the OpenBCI_GUI. (One reason being that the GUI only shows a limited period of time in the Time Series graph.)
But rather to look at other apps / frameworks which support these sort of operations. Then you could possibly customize further to achieve the desired result. I've already suggested the open source app, EDFBrowser. But one caveat is that it may need some prefiltering of the EEG recording first, to remove the DC offset (values of EEG above say 80 microvolts.) This is done by a high pass filter, for example at .5 Hz.
Another candidate app would be EEGLAB. Which runs under MATLAB.
Finally, as mentioned previously, the field of "affective BCI" and EEG emotional labeling is large and growing. I earlier posted some links to search requests that would pull up relevant papers. Your questions keep coming back to examining the EEG FFT, around events of interest. As far as I am aware, EEG FFT alone is not going to be enough analysis to categorize emotional states. Examining some of the current research papers in this area, will demonstrate the wide variety of complex signal processing required. In many cases, papers which are demonstrating emotion recognition / categorization -- also do some 'training' of ML Machine Learning algorithms based on the subjects response. The subjects in this case are exposed to standardized tests which are intended to evoke specific emotions. Then the ML system can train itself to do the recognition. FFT alone (of a single or multiple channels), is not sophisticated enough to do this type of labeling.
Comments
I'm not sure I understand your English. What is a "PPT Plot"??
FFT of EEG does not really tell you very much, although it can show which frequencies are predominant.
I already told you that normal range of EEG is up to say 80 uV. 30,000 uV is pure noise. Or perhaps you are confusing the DC offset. Please see previous link.
I don't mean to be rude, but does your supervisor have any previous exposure to EEG or digital signal processing? What are you trying to achieve by collecting the FFT at certain moments? If you are seeing large spikes in uV amplitude in your time series, that usually corresponds to noise caused by movement, muscle artifact (EMG), headset motion, etc. etc.
We trying to collect emotions based on specific events and see how they display.
It maybe me. I need to have a different scale range and I am not sure and maybe I am asking wrong questions.
GREAT. Thanks for finally explaining what you are trying to achieve (!)
The FFT does not indicate 'emotional' content. There is an entire field of neuroscience devoted to this. Try some of these search links and do some reading.
https://www.google.com/search?q=eeg+emotional+valence
https://www.google.com/search?q=affective+bci
https://www.google.com/search?as_q=affective&as_sitesearch=openbci.com
He was asking questions can we export time data (say 8 hours) in Excel. To me, it seems possible because the software does not put any limitations.
and other question was can we picl event in 8 hour recording and get FFT of event. To me, why now if it has been recorded. Maybe not in the scv, but MATLAB it should be in case it has been recorded. and FFT goes every 20 seconds.
There will be a qualified people involved for setting and analyzing experiments. and for my sake, I read everything possible because it makes me easier to prepare everything for the experiments. to me, range of 20-30,000 did not make sense and I kept calling everybody feeling my ignorance. So if I understand I could explain why it is possible or not possible. It drives me crazy because I just do not understand what should be measured. - https://fda.gov/media/74662/download
To reiterate, the FFT is not going to help you in any way whatsoever, to determine emotional valence or emotional state. By itself. Emotion recognition requires sophisticated DSP, such as is outlined in the research paper links previously given. This type of EEG research is still in early stages and DSP required is substantially sophisticated and complex.
PLEASE have your company and yourself do some reading on EEG and emotion. See previous links. You cannot do this with FFT alone. So your entire outline of what you want to do at this point, would be extremely frustrating for all involved. Putting it bluntly.
See, my boss is a scientist. I just got a job, and need to follow directions. So I can learn programming, or anything he needs from what I understand, and I am grateful for opportunity to learn. To me, those just requirements and maybe later I will understand why.
PLEASE have your company and yourself do some reading on EEG and emotion. See previous links. You cannot do this with FFT alone. So your entire outline of what you want to do at this point, would be extremely frustrating for all involved. Putting it bluntly.
Two way communications is essential in any organization. Your supervisor needs to do some background reading.
Thank you! I understand. - if there are going to be scientists involved, they can be lay audience in terms of usage of the software. So it should be as simple as possible. And it seems to be my responsibility for now and then to fill in where it is needed.
It is my job to do all the readings and explain all things regarding the software. It was my idea to purchase it because I thought it was the best on the market for the purpose.
I hope I will figure it out.
I cannot help you any further, unless you have specific OpenBCI questions. If you are interested in EEG and emotion, you need to do some background reading on how that is currently being done. FFT of noise bursts will not help you.
Thank you! I know. Maybe they want to measure the noise of aircrafts. I promise you I am doing all background reading and will do more. scale did not make sense to me. and I need to speak up and start asking questions why. So far I am focusing on what shall be done. Take care! I appreciate your help and patience.
my goal is to figure our hot to make changes to the scale from 20 to 30,000. I still have this question for me open.
EEG is generally no larger than 80 uV. Did you read my previous comments??
Yes, I did and I found this info online. I was not sure if I understand requirements.
The three types of potentials that can be measured are EEG, ECG, and EMG. The EEG only gets up to say 100 uV at the most (this is unusually high). Normally ranges from a few uV to below 80 uV. The EMG and ECG can get into millivolt range, (thousands of microvolts). 30,000 uV is 30 mV (millivolts).
https://www.google.com/search?q=range+of+emg+signal+voltage
If measuring EEG, EMG is considered noise and is usually filtered out with various means.
With your special purpose emotional detection / recording, you will be using your own signal processing. Initial stage of that will filter out EMG and eye blink artifacts. Note that the raw EEG recording contains noise + valid EEG. There is usually no need to plot the EMG noise signal.
"FFT of EEG does not really tell you very much, although it can show which frequencies are predominant."
We need to have a window with the time, so when we use PlayBack widget, we can select a time for FFT event to see what happened during that time of the event.
I have not figure out yet alternative - which program to use to pick the event and get FFT of the event.
Please, give me suggestions.
It seems more like a programmer can add a time drop down or just where a user can add the exact minutes/seconds. Any suggestions? Thank you!
Lisa, hi. I changed the title of this thread to: "collecting emotions based on specific EEG events", for more clarity.
The easiest approach for capturing the EEG + FFT around the time period of a specific event, in general will NOT be to pursue making changes in the OpenBCI_GUI. (One reason being that the GUI only shows a limited period of time in the Time Series graph.)
But rather to look at other apps / frameworks which support these sort of operations. Then you could possibly customize further to achieve the desired result. I've already suggested the open source app, EDFBrowser. But one caveat is that it may need some prefiltering of the EEG recording first, to remove the DC offset (values of EEG above say 80 microvolts.) This is done by a high pass filter, for example at .5 Hz.
Another candidate app would be EEGLAB. Which runs under MATLAB.
Finally, as mentioned previously, the field of "affective BCI" and EEG emotional labeling is large and growing. I earlier posted some links to search requests that would pull up relevant papers. Your questions keep coming back to examining the EEG FFT, around events of interest. As far as I am aware, EEG FFT alone is not going to be enough analysis to categorize emotional states. Examining some of the current research papers in this area, will demonstrate the wide variety of complex signal processing required. In many cases, papers which are demonstrating emotion recognition / categorization -- also do some 'training' of ML Machine Learning algorithms based on the subjects response. The subjects in this case are exposed to standardized tests which are intended to evoke specific emotions. Then the ML system can train itself to do the recognition. FFT alone (of a single or multiple channels), is not sophisticated enough to do this type of labeling.
Regards, William
FFT sub-window in EDF Browser, showing the FFT for the portion of EEG selected.
Thank you, William!