These electrodes are placed here...
...in the part of the brain that we know...
...controls counting backwards or moving your fingers.
Peter, you can make my brain signals audible, right?
Yes... - These are my brain signals.
Indeed. We're measuring the electrical activity with an electrode on your skull.
So it's far away from all those neurons that are firing off signals.
They allow you to stand. - It's something I transmit?
They're very weak signals. But those strong signals...
...are signals from your muscles, not your brain. Start chewing for example.
See, the entire signal is disrupted. They're disruptive signals.
So when I move my face, you get these big spikes.
Or when you blink. - Yes, indeed.
The signal is very complicated.
It's hard to make a reading of this...
...because your brain is busy and we're listening to all those signals at once.
What we've done is: We've adjusted the pitch of the signal...
...so you can actually hear it. Let me turn it on.
Oh, it sounds like a thunderstorm.
These are my brain signals? - Your neurons firing off signals.
That's what you're hearing. - My brain's electrical activity?
Yes. We can also play it in a different way...
...to make it sound a bit more pleasant.
Slowly the signals become even.
Oscillations start emerging.
Especially when you close your eyes or don't move at all.
When you talk there's a lot of activity in your brain.
You have to find the right words, and put them in the right order.
The meaning has to correspond to what you want to say...
...and speech muscles must be directed.
But the muscles throw a spanner in the works of this reading.
Yes, when I start talking it immediately starts interfering.
Those spikes are muscle movements, right?
Yes, muscle signals can be 10 times, 100 times stronger...
...than the signals we pick up from the brain.
Usually, we look for very specific activities...
...which reflect what you're doing.
For instance, imagined movements, for people who are paralysed.
They can still imagine movements.
The signal is similar to that of real movement. We can detect that.
Selective attention. For example, focusing on left or right...
...with your ears or eyes, is also a strong signal that we can detect.
This is a toy. For that you'd need a proper system.
Yes, you need a good cap and proper signal processing.
This is a gimmick, just to make it audible.
Fasten it.
I'm measuring to see if the cap is aligned.
This is a totally different kettle of fish, compared to that headset.
These electrodes give more precise readings.
We add some gel for the best possible contact.
We measure the same electrical signal, EEG.
We then try to detect what you're focussing your attention on.
We use this with paraplegics or people who can't communicate.
People with ALS, for example, in the end stage of the disease...
...who can smell, hear, feel and see, but no longer talk or move.
They become completely locked-in.
But there's still brain activity.
Detecting that activity would allow us to control something.
With this, for those kinds of people...
...we make it possible for them to type.
You focus your visual attention...
...on a place on the screen with the letter you want to select.
The letters flicker in different patterns. With that we can predict...
...what your brain signal would be if you looked at that letter.
We then compare that to the real signal...
...and conclude: You were focussing on that letter.
Backlight.
It worked. With some difficulty.
I had to use backspace a few times to get the right letter.
It reads my brain correctly. It's quite magical.
Yes, you see how reliable one must be...
...which is hard for patients who tire easily.
Ideally, you would only have to think of the letter E...
...instead of looking at a flickering matrix.
Is that possible? - Not with an EEG cap.
The signals are so weak.
But with electrode implants those signals become a lot stronger.
That would be possible. - Theoretically.
The first steps in that direction have been made.
This is a tin can throwing game.
You just said it was magical how the letters appeared on the screen.
This is a mental ball that...
A mental ball? - Yes.
It's a gimmick, a gadget.
They're all flickering now. - To explain what we do.
The computer is gathering information about the response of your brain.
Let me throw a mental ball.
Now what? - It was on target.
Indeed.
I can still...
One more.
You missed it. It hit the curtain.
The last one is always the hardest.
We're giving people with locked-in syndrome an implant.
People who can't move or talk, but only blink or move the tips of their limbs.
We want them to be able to control a computer.
This box will go here, under the skin.
These cables will go up here.
These electrodes are placed...
...in parts of the brain that we've identified beforehand with an MRI scan.
Let me show you on a brain.
This is a printed brain. The skull and the brain.
From a real MRI scan, from a real person.
We've drilled two holes.
These electrodes are placed...
...in the part of the brain that we know...
...controls counting backwards or moving your fingers.
This box reads the brain signal and amplifies it, because it's very weak.
It is then transmitted wirelessly to a receiver and computer that decodes it.
The goal of the experiment...
...is for the patient to generate a mouse-click by thinking about it.
Compared to colleagues in the US who are controlling robotic arms...
...which is a big thing there, it is...
The difference is that we want 100% accuracy.
We only want it to generate a mouse-click when the user wants it to.
Most groups have an 80% success rate, so eight times out of ten.
So two times out of ten the machine sees something you don't want...
...or it doesn't react.
We want to approach 100%.
80% is not good enough for a paraplegic.
Imagine your keyboard would be faulty every fifth letter.
Within ten minutes you'd throw it against the wall.
An implant has to be a lot better.
So what we're trying is very simple: a mouse-click.
When that works we'll move on to the next step:
Trying to decode thoughts, or rather, language.
You can talk to yourself inside you head.
That too creates patterns of activity. We have different electrodes for that.
These dense mats, in which the electrodes are located close together.
The brain is incredibly detailed.
Every square millimetre has a different function.
This allows us to recognise patterns that are connected to pronouncing a letter.
Like oo or aa. When you imagine a letter...
...we want to decode it and create a speech recognition computer.
What are you doing with this artificial head?
We're locating the precise place where the electrodes should go.
It has to be accurate to within a millimetre.
It's never been done before, so we have to develop it ourselves.
A trial operation then?
This is the second attempt. We know what to do, but we're just double-checking.
What we do is: We put people in an MRI scanner.
It's used to measure brain activity.
We've developed a new model...
...that enables us to reconstruct what people see.
For example, on the left is what people see in the scanner.
Handwritten letters.
On the right you see reconstructions based on the measured brain activity.
Isn't that going too far? You could say you can read along with someone.
Correct. But it's important to realise there's a difference...
...between observation, seeing something, and imagination, thinking of something.
The focus is on observation, but these techniques will, in the end, enable us...
...to show imagination or internal thoughts.
This is what you see, not what you think? - Correct.
The areas that process visual information...
...show a straightforward image of what happens in the external world.
It only looks sort of similar.
I see a kind of blob. - That's right, it's not perfect yet.
It's the first step in this type of research.
But we're working on ways to increase the level of detail.
Other people are looking at internal processes.
For example, at some point maybe we can read thoughts and dreams.
Perform the tasks as well as you can. Relax.
Try things out. That's the idea behind neuro-feedback.
What helps to increase the signal?
So things like: drawing faster or slower...
...or drawing something else.
Không có nhận xét nào:
Đăng nhận xét