Hello Dinesh

You will need to give us all much more information. What experimental model? Human, non-human primate,rodent etc. What equipment are you using? What is the nature of the noise you are experiencing? Electrical mains frequency (50/60 Hz), high frequency etc.  What are your filters set to?

Regards

John Reynolds 

Dear John

 I want to detect the drowsiness of the person.  The device is to be used related to Bluetooth.  Wireless EEG headband is to be used and FFT features can be used for

any probabilistic or statistical learning algorithm to see the person is in the mode of drowsiness or not.  Before this, we have to determine the probability of boundary value of the drowsiness.  How to over come the noise is the secondary thing for us now.

Thanks for your valuable opinion, suggestions and interest shown.

Best Regards

#Dinesh

EEG signal are electrical signal which are nothing but the electrical potential generated by hearth, and its strength is very low.

One needs carefully designed filters to remove noise due to artifacts and power line noise.

Hi Dinesh

OK this is more a question regarding application of EEG in humans which is not in my area of expertise. I would think a good solid literature search should turn something up for you in terms of making a criterion for change in power in specific bands ? increasing delta power for example. You obviously have a system that will send the signals to a computer that can do online FFT, so that's a good start. 

Hopefully others here will be able to help.

Best regards John.

Activity recorded from the scalp is influenced by the brain activity, muscle activity, and environmental electrical artifact. However certain components of these signals (e.g. the alpha band, 8 - 13 Hz) in EEG have been shown to be related to brain state and cognition. Please see the attached book chapter, in particular the Artifact Removal section, Section 6.

Chapter Electroencephalography (EEG): neurophysics, experimental met...

No amount of reading and thinking can replace getting into the lab and systematically varying possible sources of interference and adjusting the parameters of your recording device to reduce them. Besides, hands-on in the lab is more fun than sitting, worrying, and reading.

Bob

Dear Robert sir

 If you share your paper, just like Michael D. Nunez has done it  then that will be boon for us. No matter,  we are not so worried about data acquisition system, we are in interested to the analysis of EEG signals with their parameters and their mining of the data for getting probabilistic conclusions of the various phases of the human being.

One of theem is drowsing.

Thanks for your answer.

#Dinesh

Dear Dinesh,

From what I understand, you have already acquired the data, right? I did not quite get what is your main issue right now, is it with preprocessing steps? Have you already done some? If so, what?

Dear Carlos

Please, refer my first answer.

regards

#Dinesh

Dear Dinesh,

I appologize, but I still did not quite understand if your current problem is with noise treatment or with classifying if a given person is within the drowsiness state. Either way, I believe you could make good use of filtering your signal, in order to increase your signal-to-noise ratio, which also could help you get better features for further classification. Are you using any?

Best regards.

Dear Carlos

  Thanks for your answer.  Though we have secondary problem of removing noise.

but we have worked on it somewhat. Our main aim is to get the signal band which

can recognize the drowsiness. but it requires quality EEG signal. That' why,

we have to remove the noise.

Yes, we can use notch filter to have signal to remove noise

but still got the problem. Fs is 256hz.  

"http://in.mathworks.com/matlabcentral/answers/64949-how-can-i-remove-the-noise-from-eeg-signal?requestedDomain=www.mathworks.com"

still not getting the EEG without noise, but it is better now.

According to

"S. Haggag ; Centre for Intelligent Systems Research, Deakin University, Australia ; S. Mohamed ; A. Bhatti ; H. Haggag"

"EEG signal is one of the most important signals for diagnosing some diseases. EEG is always recorded with an amount of noise, the more noise is recorded the less quality is the EEG signal. The included noise can represent the quality of the recorded EEG signal  The method generates an automated measure to detect the noise level of the recorded EEG signal. Mel-Frequency Cepstrum Coefficient is used to represent the signals. Hidden Markov Models were used to build a classification model that classifies the EEG signals based on the noise level associated with the signal. This EEG quality assessment measure will help doctors and researchers to focus on the patterns in the signal that have high signal to noise ratio and carry more information"

We use adaptive filter and get the better result now.

Please, refer the more data attached herewith as link and pdf.

For classification of EEG signal, we refer some papers which is attached.

Ref: IEEE paper...

Classification of drowsy and controlled EEG signals

R. Upadhyay ; P. K. Kankar ; P. K. Padhy ; V. K. Gupta

#Dinesh

·  https://books.google.co.in/books?isbn=3642548792

·  ×

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3342623/

·  Undo

10.1.1.402.6015(1).pdf

·  ×

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.402.6015&rep=rep1&type=pdf

·  ×

http://liu.diva-portal.org/smash/get/diva2:746664/FULLTEXT01.pdf

·  ×

http://ijsr.net/archive/v2i10/MDUxMDEzMDE=.pdf

·  ×

http://cdn.intechweb.org/pdfs/16119.pdf

·  ×

http://in.mathworks.com/matlabcentral/answers/64949-how-can-i-remove-the-noise-from-eeg-signal?requestedDomain=www.mathworks.com

https://books.google.co.in/books?isbn=3642548792

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3342623/

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.402.6015&rep=rep1&type=pdf

http://liu.diva-portal.org/smash/get/diva2:746664/FULLTEXT01.pdf

http://cdn.intechweb.org/pdfs/16119.pdf

http://in.mathworks.com/matlabcentral/answers/64949-how-can-i-remove-the-noise-from-eeg-signal?requestedDomain=www.mathworks.com

Dear Dinesh

I see. Actually, I believe getting rid of EEG noise completely is a very difficult thing to do (not to say impossible), since there are so many environmental and physiological factors contributing to a noisy signal during data acquisition. Of course, strategies such as filtering can increase the SNR. Also, I have seen some studies using ICA or PCA to identify possible noise sources. I am attatching some of them here (I couldn't remember them all), I believe they are worth reading. Hope it can help you!

Best regards,

Carlos.

http://sccn.ucsd.edu/~scott/pdf/PSY00.pdf

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=1692128&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D1692128

http://www.sciencedirect.com/science/article/pii/S0165027015000370