BCI for Locked-in / ALS
iwoj
Vancouver
I have a family member with ALS who is gradually losing control of her muscle control. Currently, she is using eye tracking to type and communicate, but we are anticipating that she will eventually lose control of those muscles as well. I'm wondering if OpenBCI should be used to train a machine learning model to recognize some distinct thought pattern which could in turn control a Switch Control interface of a computer. All we need is a reliably responsive on/off state in her thoughts to give her control over a computer.
Is this possible? Are there products that already do this? Could we build something like this with OpenBCI?
Thanks.
Comments
Iwoj, hi.
See some of the related posts on this recent thread.
https://openbci.com/forum/index.php?p=/discussion/2404/suggestions-for-bci-addressing-dystonic-cerebral-palsy
Regards, William
Also these two searches will find P300 based spellers,
https://www.google.com/search?as_q=p300+speller&as_sitesearch=openbci.com
https://www.google.com/search?q=p300+speller
It seems that the P300 uses visual feedback to generate EEG signals, which I assume requires that the wearer have full motor control over their eyes. I'm not sure this solution will work with locked-in individuals.
What sort of thought patterns have developers been able to detect with OpenBCI? I haven't been able to find a library of examples or community applications, except for one vague reference to a remote control inflatable shark that was controlled somehow with OpenBCI.
This article says that ALS individuals still have eye control:
https://www.scientificamerican.com/article/why-do-eye-muscles-function-in-als-as-other-muscles-waste-away/
https://www.sciencedirect.com/science/article/pii/S1877065717304104
The 'inflatable shark' demo by Chip Audette, was done using multiple individuals, one for each dimension of motion. The software detected alpha generation with eyes closed. So not really a good example for you. The other thread I mentioned showed several examples of VEP: SSVEP and cVEP, both of which should work. I would suggest the cVEP would be worthwhile to get on their contact list. The other links on that post mentioned the products from g.tec, their IntendiX line.
Here is a list of various BCI paradigms, on g.tec's page,
https://www.gtec.at/product/bcisystem/
The BCI paradigms list on the previous post is a comprehensive list of types of BCI. In general, BCI does not detect "thought patterns" or mental commands. Although some systems do detect emotional valence, state of alertness, drowsiness, etc.
Here is a recent survey paper on applications of machine learning / deep learning to BCI.
https://arxiv.org/pdf/1905.04149.pdf
"A Survey on Deep Learning based Brain-Computer Interface: Recent Advances and New Frontiers"
@iwoj I understand the immediate need for systems that are both commercially available and affordable, but as William has shown, we aren't quite there with even the most advanced tech. It's certainly possible to use OpenBCI tech to build and verify an established BCI paradigm.
@iwoj
@wjcroft
Thankfully, the individual has this tech already. SSVEP requires minimal eye movement, but would require the user to look at two or more flashing objects on a computer screen.
If you are working with the Occupational Therapist, I would discuss SSVEP and cVEP. These paradigms will allow a user to make simple replies and choices. More complicated setups can allow for typing.
We are hoping that the cVEP speller from MindAffect will be priced reasonably. That seems to be their intention to get maximum outreach to the communities that can use this.
https://www.mindaffect.nl/
They appear to be attempting for CE certification for European sales. One would hope that for USA developers, this step could be bypassed. They have specifically coordinated with OpenBCI and use the Ganglion in their demos. So I'm assuming they would be open to dev kits sold here in the US. Their closed source system is provided on a Raspberry Pi 4 platform.
Thank you. I’ll reach out to them and ask. Let me know if you have any success.