How to solve the “TI ADS1299 EEG demo kit is not save for humans” problem?

[Peter Gamma]

The TI ADS1299 EEG demo kit is an affordable board for EEG data acquisition. I found two papers of Indonesian scientists who tested the accuracy of the TI ADS1299 EEG demo kit and compared it to a medical garde EEG device:
 
For a 8 channels EEG Data Aquisition System:

https://www.researchgate.net/publication/325635924_Design_of_EEG_Data_Acquisition_System_based_on_Raspberry_Pi_3_for_Acute_Ischemic_Stroke_Identification

For 32 channels EEG Data Aquisition System :

https://www.researchgate.net/publication/332321992_Development_of_Electroencephalography_EEG_Data_Acquisition_System_based_on_FPGA_PYNQ

EEG data acquisition system 32 channels based on Raspberry Pi with relative power ratio and brain symmetry index features:

https://www.researchgate.net/publication/337896178_EEG_data_acquisition_system_32_channels_based_on_Raspberry_Pi_with_relative_power_ratio_and_brain_symmetry_index_features


TI ADS1299 EEG demo kit is not a medical device and not approved for medical use. But still the accuracy data of the Indonesian scientists are very helpful.

I asked in the OpenBCI forum about OpenBCI software support for the TI ADS1299 EEG demo kit. OpenBCI supporter William Croft, said the TI EEG board is not safe for humans:

 https://openbci.com/forum/index.php?p=/discussion/2686/could-openbci-support-texas-instruments-ads1299eegfe-pdk#latest

Why is the TI ADS1299 EEG demo kit not save for humans. Here the answer from the Texas Instruments support:

https://e2e.ti.com/support/data-converters/f/73/p/940917/3476990#3476990

Is the answer of the Texas Instruments support helpful? The  TI ADS1299 EEG demo kit can be completely battery powered, and is not safe for humans? The OpenBCI EEG board is a comparable board and also completely battery powered. Why is the OpenBCI EEG board safe for humans, while the  TI ADS1299 EEG demo kit should not be save for humans?

[razvan784]

I personally think the reply from TI is very clear. It's a demo kit, an advertising tool for their integrated circuit. It's not a medical decive. It's not designed as such. It's not designed according to safety standards or principles. It's to be used with a simulator. They don't want to be legally responsible if something goes wrong. There's nothing to solve, you can't use the kit on humans and guarantee their safety under all conditions, especially component failure. It that's what other researchers did, it's their problem.
Think what happens if for example, instead of batteries, someone were to supply the board from a cheap charger that has less than 1 mm of isolation gap between the mains and the output. bigclivedotcom on youtube reviews those from time to time, it's fun watching but the idea is they're dangerous. What would happen if the power supply inside the PC connected to this board were to fail and connect 230V to the USB port? This may be a very unlikely event, but it's not impossible and the kit is not designed to be safe under such an event -- therefore is not to be used on humans and it would create a liability if it were - that's also what TI is saying.

[perieanuo]

hi,
medical electronics is a complicated subject when it come to voltage/current protections, insolation, emi.
you can't even touch the subject if you don't act in this area for 3-5 years.
I worked in radiotherapy and medical security equipment r&d, it's not as simple as it appears, you can't just take some dev board and put electrodes on a human or inject some laser or emi in a medical treatment room with arduino-competence level
the paperwork only in developed countries is a nightmare for proving that your board is ok for this matter.that's why is not so simple to draw this line, even TI avoid to decide it's safe, and they're right
so, 'safe for human' is only for those inside this r&d environment.I prefer to hear guys that worked in the real life developing this type of equipment or at least someone who has some years repairing it
regards,pierre

[Peter Gamma]

These arguments cannot convince me either, why I should buy a OpenBCI board instead of a TI EEG board, which costs half the price of the OpenBCI board.

Are the Texas Instruments board and the OpenBCI board not very similar? The Texas Instruments board is very basic, the OpenBCI board has more features, which I don t use.

Both boards are based on the TI ADS1299 chip. Therefore, data of the paper "comparison of Cyton with medical grade g.tec devices" should apply for both boards, as the paper says:


https://arxiv.org/pdf/1606.02438.pdf

William Croft argues:

"Do you understand that open-source hardware projects, are only supported by the revenues from hardware products sold? There are no other income sources. Open source hardware is very different than open source software. As there are large and significant overhead costs for hardware projects."

But what can I do, when there are on other income sources for OpenBCI, than revenues for hardware products?

The community of EEG device users is small, the biggest part of it is probably OpenBCI. OpenBCI has a support forum for EEG devices which is bigger than that of Texas Instruments for EEG.

What about this solution: Those who need support in every single step, choose OpenBCI, those who need only support for the basics, choose Texas Instruments EEG.OpenBCI users have to pay for this extended support twice the price.   

[Peter Gamma]

I don t wan t to make a medical approved device out of the TI EEG board. I need the EEG equipement for a research project.

[ralphrmartin]

Assuming your research involves human (or even animal) subjects, you will almost certainly need to get approval from an ethics committee before your research can go ahead. It seems very unlikely that a responsible ethics committee would approve an experiment using equipment which states that it is not intended for direct interface with a patient, and is only for evaluation with a simulator.

The point is that the device has not been designed with safety in mind. It has been designed for component evaluation. It may ostensibly be safe, but the designers may have overlooked something. This is one of the reasons real medical equipment is costly. Designs must be validated, and approved by safety panels, and so on.

[janoc]

I don t wan t to make a medical approved device out of the TI EEG board. I need the EEG equipement for a research project.

Feel free to buy the TI board. Just don't use it on humans.

You didn't say what you want to measure, perhaps for capturing EEG of a cow or a dog this would be fine.

OTOH, for human research you are putting your job on the line if you decide to "save" money in this way. And if anything goes wrong, it will be prison time for you. But hey, you are an adult, you are supposed to be able to make your own decisions.

[razvan784]

These arguments cannot convince me either, why I should buy a OpenBCI board instead of a TI EEG board, which costs half the price of the OpenBCI board.

You shouldn't buy either one!
Quote from the OpenBCI website https://shop.openbci.com/collections/frontpage/products/cyton-biosensing-board-8-channel?variant=38958638542, emphasis mine:

Important Notice: OpenBCI, Inc. provides the enclosed product(s) under the following conditions: This evaluation board/kit is intended for use for ENGINEERING DEVELOPMENT, DEMONSTRATION, OR EVALUATION PURPOSES ONLY and is not considered by OpenBCI, Inc. to be a finished end-product fit for general consumer use. Persons handling the product(s) must have electronics training and observe good engineering practice standards. As such, the goods being provided are not intended to be complete in terms of required design-, marketing-, and/or manufacturing-related protective considerations, including product safety

It's basically the exact thing the Texas Instruments representative told you - there's nothing physically stopping you from using the product as you describe, but you shouldn't. If you do howerver, you do so at your own risk.

[SilverSolder]

Isn't it hard to imagine anything going wrong if the device is battery powered, and not connected to anything else?  (assuming 9V battery or less)

[aristarchus]

These arguments cannot convince me either, why I should buy a OpenBCI board instead of a TI EEG board, which costs half the price of the OpenBCI board.
..

Dont know why you keep refering to OpenBCI, but from what I understand even theirs is not medical approved according to their own 'Liability' clause.

"OpenBCI products are not medical, clinical, or diagnostic devices nor are they intended for medical diagnosis and are provided to you "as is," and we make no express or implied warranties whatsoever with respect to its functionality, operability, or use, including, without limitation, any implied warranties, fitness for a particular purpose, or infringement."

https://docs.openbci.com/docs/08FAQ/Liability

[Peter Gamma]

Nobody could convince me that these battery powered devices are risky to use. If there is a risk, it is one which I can take.

 Also with Scott Hardens sound card ecg there was a discussion:



“Hi Scott , just a little word of caution. The ground pin of the sound card is probably connected to the mains earth. This could be a source of danger in case of some devices which have a ground fault. Hence it is probably much more safer to use a battery powered laptop. Of course the best would be to use a wireless Bluetooth device, i guess.
Cheers and hope you are in the best of health !

Hi @mumbai, I hear this comment frequently. The input impedance between the AD8232 and the pads is extremely high. Even if the AD8232 were grounded (e.g., through the sound card), risk is eliminated because the person is not connected to the AD8232 through a low impedance path, so a ground fault is not a concern.

Interestingly, grounding yourself to your computer is common. Many people physically ground themselves to their computer with a wrist strap connected to their computer chassis when they work inside their computer. Since that is a low impedance path to ground, a wrist ground strap strikes me as much more of a concern for ground faults than a setup like this. It's an interesting topic to ponder!

@Scott Harden  Thanks for the reply. A ground strap has a 1Meg resistance to ground, hence it is not a " low impedance path to ground", quite the opposite actually. Also when working on mains equipment always unplug the device , else/and use suitably rated probes and test equipment. The last thing you want in such circumstances is a  "low impedance path to ground"
"risk is eliminated because the person is not connected to the AD8232 through a low impedance path"
Actually isn't the conductive gel drastically reducing this impedance ? I could not get the circuit diagram of your module, so am not too sure about this.
Indeed it is an interesting topic to ponder.

[Bassman59]

Nobody could convince me that these battery powered devices are risky to use. If there is a risk, it is one which I can take.

If you are hooking yourself up to this thing, then sure, you can take the risk. Don't expect to be able to sue for damages if there is a harmful failure.

As for using it on other people? What the others have said about that, especially in the areas of human-subject testing, is spot on, and you open yourself up to all sorts of bad consequences should something, ANYTHING, go wrong. Because if that happens, even though this device isn't the cause, that doesn't matter -- you were using something not approved for medical use/human subject use and the lawyers will staple your balls to the floor, and then they'll REALLY get to work.

Don't say you weren't warned, and oh, by the way, this thread is proof of the fact that you were warned but chose to ignore the warnings.

[SilverSolder]

If you keep everything battery powered, including the laptop that is processing the data,  I would like to hear anyone explain how that could possibly pose a risk from a technical / engineering perspective.

[TurboTom]

It's not only the safety to use the instrument but also the "suitability for the job". If a physician bases a treatment on an ECG, he's got to be sure that the figures the ECG displays are correct.

Moreover, also low DC voltages applied to electrodes attached to the skin, can, if present over long time, cause serious harm, including death, to the patient.

I've once been involved in designing a medical diagnostics device prototype and I had to go through a lot of paperwork and do a lot of documentation...

[Someone]

If you keep everything battery powered, including the laptop that is processing the data,  I would like to hear anyone explain how that could possibly pose a risk from a technical / engineering perspective.

Hello Dunning-Kruger?

This has come up before
https://www.eevblog.com/forum/projects/leakage-current-problem-in-emg-circuit/
Explain how hypothetical device under single fault condition remains safe.

[Peter Gamma]

@ ralphrmartin

Thank you for your answers, this is a good point. Since there is a high agreement that the risk with completely battery powered EEG or ECG devices connected to a battery powered PC is low to damage humans or animals, how has someone to proceed to get a device which is validated for research on humans and animals?

[janoc]

@ ralphrmartin

Thank you for your answers, this is a good point. Since there is a high agreement that the risk with completely battery powered EEG or ECG devices connected to a battery powered PC is low to damage humans or animals, how has someone to proceed to get a device which is validated for research on humans and animals?

So you are not really looking for advice from people who actually know something about the subject, only for confirmation of your (incorrect) preconceived idea. And then are selectively picking evidence that suits you and disregarding the rest.

Dude, I really really wouldn't want to be your research subject if this is the way you are conducting your experiments!

[Peter Gamma]

Everybody is free to give advice or not to give advice in the EEVblog, following the slogan "no script, no fear, all opinion". If someone can convince me, I change my mind.

[Zorc]

@ ralphrmartin

How has someone to proceed to get a device which is validated for research on humans and animals?

Just buy the real deal 
There are plenty of options & companies to buy fully medical certified ECG/EEG solutions. You don't want to write any medical research paper about picking some electronic parts and soldering it together, hook up some wires and clamps on someone and presenting the results anyway....
Sure if you want to do some hobby (research) experiments on yourself, you can do whatever you like so it won't be of any real scientific value.

[Peter Gamma]

I make the real deal here in the EEVblog, and publish my knowledge about these devices here for free. Maybe someone else gives here something for free, too.

Do i need security tests at all for battery powered devices to use them for research? Is it possible to negotiate a special disclaimer for this purpose to decline liability, for instance when someone uses for instance a TI EEG board and connects it to AC during research experiments?

[2N3055]

Everybody is free to give advice or not to give advice in the EEVblog, following the slogan "no script, no fear, all opinion". If someone can convince me, I change my mind.

Do you consider 9V battery a high or low voltage?

Now go, get one and stick it to your tongue. And hold....
And then get back to me and tell me if that is OK to apply to someone or not.

Attaching electrodes with electrolyte gel to body produces similar impedances....

That is why medical devices have to have specific failure modes, that  in event of every possible failure, you can DEMONSTRATE and PROVE it cannot produce dangerous currents to the body.
Together with proving it cannot cause any malfunctions to other medical and medical assistance devices (like pacemakers, neural stimulators and such) that patients might have.

Also you misunderstood: what people is telling you is a FACT.  It is a fruit, a body of knowledge, of 50+ years of medical electronics industry. Nobody have to prove anything to you. It is already proven fact within the industry.

It is you who need to learn many things before you can even understand why is medical electronics in a current state of the art. And once you become expert in the field, you can develop new technologies that will push boundaries of performance, price etc...

It's not a conspiracy theory. Like it or not, by reading your answers, it simply seems that it is simply too complicated for you to understand, at your current level of knowledge.
Which is OK, nobody was born with PhD. You simply should start reading and learning about it.

[SilverSolder]

If you keep everything battery powered, including the laptop that is processing the data,  I would like to hear anyone explain how that could possibly pose a risk from a technical / engineering perspective.

Hello Dunning-Kruger?

This has come up before
https://www.eevblog.com/forum/projects/leakage-current-problem-in-emg-circuit/
Explain how hypothetical device under single fault condition remains safe.

It is the opposite of Dunning-Kruger to ask questions! 

The lethal current limit appears to be quite a bit higher than the 0.5mA quoted as safe for skin contact, when we are talking electrical safety in general.  Wonder why that is?

In principle, that very low limit means that even the output of a AAA battery is not considered safe to apply to skin.  That sounds very conservative - are there really examples of people who have been seriously injured by current from a 1.5V battery?

[Peter Gamma]

Not to my knowledge. But my knowledge in this field is limited, I admit. Maybe one could start a research project to find out about this?

What I found out that Chinese sellers for EEG brainwave modules are quiet responsive to discussions in the OpenBCI forum and the EEVblog. Since I started questions regarding this topic, a new product appeared on Aliexpress with an instruction not to use an AC powered laptop in combination with EEG sensors, but only a battery powered. Before I started asking question, there was no such product with this instruction there. The Chinese seem to follow the safety discussion with interest, and immediately adapt their products to new customer needs:

https://www.aliexpress.com/item/32969264474.html?spm=a2g0o.productlist.0.0.2d4e77f5NMc1lj&algo_pvid=84102863-692f-4544-bb07-6fb5da865f91&algo_expid=84102863-692f-4544-bb07-6fb5da865f91-0&btsid=0bb0624516006139479535232edbf6&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_

[DrG]

@Peter Gamma

Many of the responses, in my view, make good points.

The manufacturer of the board(s) that you mentioned at first has (have) outright said that is not for human use. They are not obligated to tell you why it is not approved for human use and they are certainly not obligated to explain to you what you have to do to make it approved, or safe, for human use.

You have said that you are not trying to market a medical device, but instead, want to use the device in research. You do not want to state exactly what you want to do and with what subjects. You have not even said outright that you want to use the device with humans. You are not being forthright.

Written in one of your references is:
“Even though medical grade equipment possesses certification and still outperforms the OpenBCI board in terms of classification, the latter gives very close EEG readings”

Spend some time and effort to understand what goes into human-use approval for an EEG device for research or for marketing as a medical device. Learn what certification means in this context.

Go to the regulators in your country or in any other country that you want to conduct research with humans with your device and see what is required. EEG devices for use in humans get FDA (US) approval all the time. You can see the approval letters online and sometimes devices are approved based, in large part, because they duplicate designs in existing and approved devices. More importantly, you can see the regulations online (although you may have to look hard and long) and you can bring your design and purpose to the attention of the regulating agency and ask for an opinion / approval – why not?

Learn the details of medical grade, consumer grade and research grade (even if those terms have little legal meanings). Again, have you tried asking regulators in your country – or even published electrophysiologists in Universities in your country?

Think about it – who wants to tell you that it is safe for use with humans after the manufacturer(s) has (have) already stated that it is not intended for use in humans?

Who here wants to be that irresponsible and if they were, why would you accept that?

[SilverSolder]

I found this chart that looks kind of interesting in this context:




Source: https://www.asc.ohio-state.edu/physics/p616/safety/fatal_current.html