Examination project of EEG, Opinions and feedback

[Farhad-.-]

Hello!

I don't really know how High School ends in other countries but in Sweden it ends with an examination project depending on the programme you've selected. Anyhow we were free to choose one research topic that is related to mathematics, physics, biology and chemistry. I choose the brain and is tied to physics/Biology.

I already bought a EEG equipment, Neurosky Mindwave mobile +. I already have the data for 8 weeks, 8 hours daily. Examination projects name is "Brain signal processing and algorithmic predictive model and applications", what I am basically doing is create a Predictive algorithm that records a users data for a few days and predicts the users state of mind without the EEG equipment and proceed to suggest different stimulus like music or coffee. There will be a testing phase where the user chooses a few different songs and record the affect of it. The predictive algorithm will notify the user to listen to a certain music to stabilize the mood of the user.

Here is the unfinished abstract

The brain constantly generates different kinds of signals. The signals are gathered with an Electroencephalogram (EEG), the signals can interpret the different mental states of the user. Data points was collected using OpenVibe and interpreted by MATLAB. The user had the EEG equipped during school for 8 weeks whereof the user had two instructions, one not to listen to music during school hours however only for 4 weeks. The user was also instructed to listen to music the remaining 4 weeks. Music is a stimulus and would affect the users state of mind. The second instruction was to write down the state of mind every two hours, approximately four times during school.

The reason I made this topic is to get your opinions, feedback and any sort of improvement you might have and tell me if anythings unclear.

Thanks for reading :)

 

[berkeman]

Fun project.

How did you deal with movement artifacts of all that movement while acquiring the EEG data?

 

[Farhad-.-]

Fun project.

How did you deal with movement artifacts of all that movement while acquiring the EEG data?

Well the EEG machine I am using has a built in feature as in an ear clipp that acts a ground and reference, which in turn focuses on the brain.

 

[Fra]

Artifact handling, and experimental control are the most difficult parts. In what you describe it sounds like exact experimental control is not possible, but perhaps you can get away with assuming that over long times, you can still extract some average modes. But i think one should probably investigate or have some theory that makes sure that the underlying effects you seek, will as per your paradigm dominate over the uncontrolled conditions.

Well the EEG machine I am using has a built in feature as in an ear clipp that acts a ground and reference, which in turn focuses on the brain.

A ground electrode does not eliminate artifacts.What EEG does is just to record differences in scalp potentials, usually between various positions(Vin+) and a reference(Vin-). The GND is usually separate from the reference unless the conicide.

These scalp potentials are contaminated from many things beyond the brain activity. For example eyeball movements (the eyeballs are dipoles that give a relatively strong DC field), EMG such as eyeblinks,mechanical movements, causes electrodes and skin to vibrate which pulls leads, swalloing, chewing, sneezing, sometimes even the electrical field of the cardica muscle can give measurable effect in EEG. This is because the electrical field of the heart and EMG are STRONG relative to the very weak fields of EEG.

surface EEG is usually the order of 0.2-200uV, EOG, EMG and ECG are usually way larger

There are various methods all from simple frequency filtering, to advanced algorithms that attempts to clean the EEG signal from atypical components, if you measure EOG and eyeblink EMG you can even do independent component analysis and mathematically separate sources.

But one way or the other you should at least be aware of this. For sure, the ground electrode on the ear does not solve the above problems. Ie. you need to be able to argue that whatever your results are, are not biased by the lack to control of conditions and artifacts.

Unless you already as part of pre-research checked this, one of the first google this for example gives this
https://www.slideshare.net/ranjithpolusani/artifacts-in-eeg-final but you can

This sounds like a fun project though. No doubt that stimulus like audio (music) affects the state of mind. I know a few researcherin Sweden that actively work on music psychology. So if you want contact with someone to assist, check this out
http://www.psyk.uu.se/research/research-groups/the-music-psychology-group/

/Fredrik

 

[Farhad-.-]

Artifact handling, and experimental control are the most difficult parts. In what you describe it sounds like exact experimental control is not possible, but perhaps you can get away with assuming that over long times, you can still extract some average modes. But i think one should probably investigate or have some theory that makes sure that the underlying effects you seek, will as per your paradigm dominate over the uncontrolled conditions.

A ground electrode does not eliminate artifacts.What EEG does is just to record differences in scalp potentials, usually between various positions(Vin+) and a reference(Vin-). The GND is usually separate from the reference unless the conicide.

These scalp potentials are contaminated from many things beyond the brain activity. For example eyeball movements (the eyeballs are dipoles that give a relatively strong DC field), EMG such as eyeblinks,mechanical movements, causes electrodes and skin to vibrate which pulls leads, swalloing, chewing, sneezing, sometimes even the electrical field of the cardica muscle can give measurable effect in EEG. This is because the electrical field of the heart and EMG are STRONG relative to the very weak fields of EEG.

surface EEG is usually the order of 0.2-200uV, EOG, EMG and ECG are usually way larger

There are various methods all from simple frequency filtering, to advanced algorithms that attempts to clean the EEG signal from atypical components, if you measure EOG and eyeblink EMG you can even do independent component analysis and mathematically separate sources.

But one way or the other you should at least be aware of this. For sure, the ground electrode on the ear does not solve the above problems. Ie. you need to be able to argue that whatever your results are, are not biased by the lack to control of conditions and artifacts.

Unless you already as part of pre-research checked this, one of the first google this for example gives this
https://www.slideshare.net/ranjithpolusani/artifacts-in-eeg-final but you can

This sounds like a fun project though. No doubt that stimulus like audio (music) affects the state of mind. I know a few researcherin Sweden that actively work on music psychology. So if you want contact with someone to assist, check this out
http://www.psyk.uu.se/research/research-groups/the-music-psychology-group/

/Fredrik

Thanks, I'll certainly keep everything in mind :)