Need HELP in choice

Pages: 1
Need HELP in choice, researcher version EPOC-EEG
Hi, I intend to buy this device (Research version) and want to ask a few questions.
1) what is the sampling rate of the signal in Hz?
2) is it possible to change the location of the electrodes. for example, to concentrate them around the motor cortex or they are firmly fixed?
3) are these electrodes equipped with preamps?

i am really interested in EEG study and want to use motor imagery based BCI.

please answer owners of this unit.
I really dont want to be mistaken with the purchase.
Hi Dmitry
I'm trying to answer you again (see my previous posting) and hope this works. I've taken the precaution of recording it as a Word document, so I have it in case this doesn't go through.

1. The sampling rate is 128 kHz

2. The electrodes are fixed in location but there are two dummy electrodes which act as pivot points on the mastoid process just behind the ears. These can be exchanged with the reference electrodes over the P1, P2 area and folks apparently do this for various reasons, although I’ve never done so. I would have liked to have at least one sensor on the central scalp area, like CZ, which I’ve found a bit of a shortcoming for various tests that I’m doing, but can live with this. The device also contains a gyro detector to detect head movement.

3. The headset is self-contained and communicates with the computer via a wireless-connected “dongle” which plugs into a USB port and works very well.

4. If you want to do serious research you must get the research SDK, but this is limited and only give the basic EEG plotting facilities. You really need professional software like EEGLAB, which I use, for more detailed analyses.

5. The device is not a Mickey Mouse gaming setup, and has been proven to give professional results. Obviously you get what you pay for and you cannot compare it with a professional medical setup costing ten times more – but there have been a few professional papers which have established that for the price you are getting excellent value, and a device which offers a lot of potential as a BCI interface for professional research.

Here are some of my findings:

1. The biggest problem with this device is to position it correctly so that it makes perfect electrical contact with the scalp. This can be very frustrating, particularly when positioning it over a woman’s head with a lot of hair. It takes quite a bit of juggling but once in place is pretty stable.

2. A weakness of the device is that if you over-stretch it, it can crack, so you have to be very cautious in putting it on a subject with a large head. There are quite a few postings on this forum. The most valuable one that I have read, is by Robert Oschler, a few postings down, who , after cracking his headset, resorted to the use of one of those simple and cheap “Alice Band” hair clips, which fits perfectly over the headset, and can be easily taped to it. I have done this using simple masking tape, and the results are excellent. Not only does it enable the headset to grip much more snugly, but it also improves the electrical contact and reduces setup time significantly. Something to seriously consider. Also, I advise you to use plenty of saline, which is available at any pharmacist as contact lens solution.

3. Another problem is that the little electrodes do not always fit in securely, although they are supposed to clip in easily. I found that this is not so, and consider it a weakness of the device. You have to just be very careful, because its easy to lose these electrodes. In particular the felts obviously wear out, and for hygienic reasons as well, I definitely recommend that you purchase a pack of spare felts, as well as a spare “hydrator pack” which is a set of spare electrodes.
I am using this device successfully for research purposes and find that I am getting excellent results, considering its limitations.
Hope this helps.

Stan, thanks a lot! thanks for your review. We must once again think twice before purchasing this device. I extremely do not like the fact that the electrodes can not change their position. I also dont think, it would be acceptable to break device for our needs.
You are right when you talk about professional software and I use eeglab too, but as you know only for offline analysis. Online BCI sessions required telemetry device, and we assumed that Emotiv is a good candidate for this role.
The area of consumer BCI has three major players: NeuroSky, Emotiv, and OCZ. Emotiv EEG has significantly more electrodes than its competitors, and slightly more expensive. In addition, we need ergonomic headsets with quick installation.
I also learned a little publication associated with the device and convinced of its suitability, you can also use the open vibe platform
Here a reference:
Thanks for those references Dmitry. Yes indeed, Emotiv CAN work if you are using it within its capabilities and not expecting to get the results of an expensive medical setup. The biomedical site is pretty harsh on it, but again, they're looking at things fr om a medical application point of view. For me, I'm using it to identify late latency components like MMN's and it's doing just fine. If your issue is to be able to have more versatility in electrode placement, well, its going to be lim iting. I wish you luck.
Stan, course, we'll try to do something out with the installation of the electrodes.
i think this problem potentially will solved.
I also wish you success and valuable discoveries!
Thank you once again, Dmitry.
Hi guys! Thanks Stan for filling the gap - I've been otherwise occupied for the last few weeks.
Regarding the Biomedical Engineering reference, we have not responded as yet but we believe there are several inaccuracies and overstatememnts of fact in the article. For example, the cost of ownership of EPOC is higher than expensive research model EEG systems? This is just ridiculous. Apart from any proper accountancy questions related to amortising the cost of a $50k device against a $750 device, EPOC basically runs on salt water - you can make your own fluid virtually for nothing. Properly cared for, you don't need new sensors, but of course a few spares won't hurt. A couple of spare sensor kits and a pack of spare felts and for $1000 yo are set up with nothing else to buy. Conventional EEG systems either use gel (not cheap - around 20c per headset application by our calculations) or disposable electrodes (about $15 for a set of 16). Not to mention the 30 minutes to 2 hours of technician time to fit the gear. Anyway... the technical comments are equally unfair. EPOC has many validation studies independent of that paper. Please see the full list <here>
I just completed a major study collecting ambulatory and very active EEG in a quite challenging location where we developed a specific detection requested by a customer. The EPOC system performed flawlessly and I was able to collect about 30 hours of clear, usable EEG from 18 subjects of different ages, genders and backgrounds - and of course with varying head shapes, sizes and hair cover. We did not need to throw away any data at all - simple care and attention to proper EEG setup meant we were able to collect over an hour of decent EEG from each user session as the user went about their normal activities including some significant movements. The task was completed to schedule and we will be publishing results after the client lifts their embargo.
The only 'shortcomings' of EPOC are related to the fixed sampling rate (effective bandwidth is 0.2-43Hz), which precludes ERP studies faster than about 50ms, and the 'fixed' locations of the sensors. The data is cleaned effectively in the preprocessing stage (down sampled from 2Kz per channel) so apart from super-fast synch requirements the low sampling rate is not a problem for most studies. The 'fixed' locations can be worked around readily - for example by tilting or reversing the headset to access most positions, although there is limited coverage of the central line (we have a processing solution to this question - using source localisation we can recover the effective central locations, but of course this is modelled rather than directly measured). Or you can take the approach of Professor Debner (see our reference list) and use the EPOC electronics with a conventional cap - this gives you a fully supported clean EEG data stream through a 14 channel wireless amplifier system which can feed directly into OpenViBe, BCI2000 or directly into Matlab/Simulink or may other supported paths.
Pages: 1