I am a computer science student, and as part of my project, I would like to develop a system that changes the TV channel, increases its volume, etc. by just the thought of it. My primary investigation paved the way to EEG BCI.

Which fields should I research in order to develop such a system, and what equipment in general is necessary? As far as I know we need an EEG device attached to head, a transmitter to send the signal to the TV, a receiver at the TV side, and proper processing of signals and feeding to the TV. I mostly need more explanation of the EEG side.

  • $\begingroup$ All of these questions were answered in your previous question. You need to ask more specific questions. $\endgroup$ – Louis Thibault Jun 7 '14 at 15:26

Beyond what you've already listed, you'll need advanced signal processing skills, as so far, nobody has figured out how to get much more meaningful information out of the EEG than broad attentional state (alpha blocking and the P300, and the evoked potentials), and maybe some correlates of motor imagery, though none of these have realistically proved themselves yet.

You will need to clean the signal using techniques such as independent component analysis/ICA and wavelet denoising, and identify some weak correlate of mental activity in the strong irrelevant "background" activity. Approaches regarding the latter fall into two camps, as I've noted above: either try to identify some well-known neuropsychological component, such as the P300, on a single-trial basis; and/or use machine learning to agnostically identify a signal, with Support Vector Machines being en vogue right now as far as I know. I assume the latter part you're already informed about.

Check out MOBILab for some of the leading figures with regards to EEG signal processing, but be aware that the EEG is an extremely noisy measure.

  • $\begingroup$ Do you have any suggested references on the SVM or other machine learning techniques with EEG signals? $\endgroup$ – Josh de Leeuw Jun 8 '14 at 1:34
  • $\begingroup$ No, I'm sorry. Nothing I'm personally even halfway convinced by. I have very little faith in what I've seen so far. $\endgroup$ – jona Jun 8 '14 at 1:40
  • 1
    $\begingroup$ onlinelibrary.wiley.com/doi/10.1111/j.1469-8986.2012.01471.x/… Read this though for some sweet LDA. $\endgroup$ – jona Jun 8 '14 at 1:40
  • 1
    $\begingroup$ @Josh A colleague of mine recently defended his thesis, which deals with decoding MEG signals in real-time. I think this may offer some interesting approaches to feature extraction and classification that may or may not lend themselves to BCI. $\endgroup$ – Louis Thibault Jun 8 '14 at 14:51

On the EEG side, you'll need a way to access the EEG data in real-time so that you can perform whatever computational techniques you need to do in order to generate signals to control the TV. Some of the newer commercial-oriented EEG headsets have SDKs that you may be able to use. We've used the Emotiv system in our lab for mobile EEG research, and they have an SDK that can be purchased for parsing the raw EEG data.

The processing pipeline will probably be something like: EEG -> Computer that is processing the EEG signal -> TV Transmitter -> TV.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.