4 channels on the ganglion, yet room for 61 sampling locations - why?

Taboen

I'm new to OpenBCI and BCI in general. What is not clear to me is why there are only 4 channels on the ganglion and yet the Ultracortex Mark IV headset has 61 locations. This seems not logical. Why is this? Can multiple locations be measured through one channel? 

Kind regards

wjcroft

https://www.google.com/search?q=10-10+eeg

The 10-10 and 10-20 electrode placement systems are just landmarks for placing your electrodes. These same positions are also available on other headset systems, as this is an international standard.

Electrode placements and the number of channels supported by the EEG amplifier are separate issues. With the mounting arrangement on the Ultracortex and the OpenBCI boards, 8 or 16 channels can be supported. And 4 (or possibly more) with the Ganglion.

By running cables to another type of amplifier, more channels could be supported. Many neurofeedback (and even some BCI) applications can be realized with 4 or less channels. High channel counts are more important if you are doing 3D source localization, or QEEG, etc. Clinical and research systems can have 64, 128 or 256 channels. Which cost many tens of thousands of dollars.

http://www.egi.com

Taboen

Thanks for your response. I want to know whether consumer type devices (relatively cheap, <500 dollars) can aid in mental exercises for a better attention span. 

My motivation for this is that I have noticed that in my daily life my attention works very different when I meditate daily compared to when I don't. Meditation improves attention duration and ability to focus on and comprehend information. As a side effect is also lowers my reaction times. I don't yet understand what kind of brain activity this is related to and I want to see whether I can make an exercise in which the computer gives feedback. Sometimes meditation works much better than other times, and it feels like I am poking in the dark. I am wondering whether measuring EEG could possibly shed light on this matter.

How many channels do you think is suitable for this type of EEG use?

wjcroft

The neurofeedback that is done with ADD / ADHD depends a lot on what the brain map looks like. This is called a QEEG, quantitative EEG, measured typically at all 19 10-20 channels. Then compared to a normative database. The actual neurofeedback sessions then are typically run with 2 to 4 channels for traditional style neurofeedback. So yes, a lot can be done with just 2 channels.

Frequently attention deficits can correlate with larger than 'normal' amounts of 'slow' waves present in frontal regions. Slow waves being considered those in the range of theta, alpha or delta. Some training protocols reward for decreasing what is called the theta/beta ratio, to try to reduce excess theta. Or strengthen the ability to 'inhibit' excess theta over time.

Checkout Pete Van Deusen's http://brain-trainer.com website, this is a low cost way to do home training, including the initial assessment. Another way to go is with the inexpensive Muse headset and apps; though this is just measuring forehead points, and not working with the entire 10-20 system.

Meditation is always an excellent adjunct to any other modalities you are using. It has numerous benefits. However some styles of meditation do reduce beta activity or increase alpha or theta, which might correlate with the reduced reaction times you mention. In other words, cultivating a serine monk-like state of mindfulness and calm, might be incompatible with a kind of twitchy fast reaction time needed in FPS games(!)  :-)

William

PS Here's some Muse info,

https://www.google.com/search?q=muse+eeg+focused+attention+training

Taboen

Thanks for your detailed response. 

Do you yourself think EEG is useful at all for complementing a meditative practice?

wjcroft

I get the impression that a number of companies (Muse included) are looking to develop technology that aids in meditation practice. At the same time, we're just in the very early stages of this exploration.

So primarily I would connect with some good teachers of meditation practices that resonate with you. The tech can then be used as kind of a progress indicator or way to track brain changes over time.

Another (bit more expensive) headset is the Versus,

https://getversus.com/

wjcroft

Recent editorial / paper on meditation research,

http://journal.frontiersin.org/article/10.3389/fpsyg.2015.01731/full

Editorial: What can Neuroscience Learn from Contemplative Practices?

... Significant progress has been made in the area of the neuroimaging of
meditation and mindfulness, leading to increased understanding of the
neural mechanisms underlying different techniques and stages of
meditation (Lutz et al., 2008; Travis and Shear, 2010; Vago and Silbersweig, 2012; Craigmyle, 2013; Josipovic, 2014; Tang et al., 2015a).
Results point to increased flexibility and efficiency of the brain's
networks, and to enhanced functional and structural integration among
their nodes (Braboszcz et al., 2013; Luders et al., 2013; Tang et al., 2015a). ...

BucketOfFish

I notice that the old OpenBCI 32 bit board was able to read out 8 channels. Why did this new iteration drop to 4 channels? What is the cost involved in creating boards with more readout channels?

wjcroft

The V3 chipKIT board uses the Texas Instruments ADS1299. It will continue to be available.

The Ganglion is a much lower cost solution, using a different chipset with fewer channels.

BucketOfFish

wjcroft, do you know how much benefit can be gained from increasing the number of electrode channels? You mention that having a large number of channels is mainly good for performing 3D localization, and that a lot of applications can be achieved with only a small number of electrodes. However, wouldn't better 3D localization allow for better removal of things like EMG signals, which are a large impediment to clean EEG readout? That is, if you can tell that a signal is really coming from say the visual cortex, rather than the face, then that seems like that would be extremely useful in actual applications.

I'm not saying that a 100-channel system is feasible commercially, but I'm just wondering how big of a difference the channel number actually makes. Obviously right now we don't have any systems really capable of responding to brain signals in real time, at least not to the degree where you could e.g. move a robot hand as though it were your own. Would you say the limitation is channel number, measurement noise, or instead something more intrinsic, like the attenuation of signal through the skull? What would you say the bottleneck is?

wjcroft

BOF, hi.

Much neurofeedback is done with only 2 or 4 channels. The neurofeedback I do is with 4 or less channels. All of these respond in real time.

Source localization may eventually take hold in the consumer / Maker market, but the current weak link is in dry headsets with even coverage of the head. And 19 or more channels. Source localization IS widely in use already in research, clinical, commercial applications; those markets that can afford it.

BCI systems respond in real time. Look around for the threads on the OpenViBE P300 speller and motor imagery. These are not using source localization but rather training on feature extraction. They are also running with a small number of channels, 8 or 16. See the OpenViBE site for more examples.

There are existing BCIs controlling robot hands done with cEEG cortical arrays implanted beneath the skull. Dean Kamen (DEKA / DARPA) is working in this area.

See this post for some links on eye blink artifact removal, without large numbers of channels or source localization,

http://openbci.com/forum/index.php?p=/discussion/comment/805/#Comment_805

BucketOfFish

Hi wjcroft,

Thanks for the response. I am aware of techniques such as P300 and SSEVP, but I consider those techniques to be somewhat clunky, because you have to keep your attention on a flashing screen, and you can only read out a small number of commands per second. By real time response, I was referring to the fact that in higher-resolution imaging techniques such as fMRI, a researcher can see exactly which parts of a patient's brain light up when the patient thinks about different concepts such as e.g. moving a left arm versus moving a right arm. If EEG had similar resolution, a user could think about moving left or right directly when moving a quadcopter, for example, instead of having to look at a flashing left or right arrow.

Could you point me to a good paper where source localization has been used? I want to see what the technology is capable of. And if I understand correctly, you say that the current limitation for consumers is a headset with a large number of channels? What would be the difficulties involved in taking the OpenBCI headset and just adding more data acquisition boards? Naively, if I just had five or six of them stacked on top of each other, couldn't I read out that many more signals?

wjcroft

https://www.google.com/search?q=eeg+source+localization+review

The current V3 chipKIT board has a serial port data thruput issue between the main chipKIT processor and the onboard RFduino. This is currently limiting the number of channels that can be stacked. Several other threads on the forum discuss this, use the search box.