So it's a great pleasure to
have him with speaker today.

That's his degree if I was in
the early de Broglie master's degree.

You know let me see if you know this
In Washington University of Michigan and

in between there she also had
a number of students industry.

Michael Ware Motorola and
eventually assistant

professor of computer engineering here in
Georgia Tech in two thousand and seven.

She also holds an engine
rebuild recent memory

and she actually got three of us

agree in clinical inference we're still
research from two thousand and thirteen.

She said career or been a real
pleasure to have well thank you.

I appreciate that introduction.

It's it's a pleasure to talk
about the work that my research

group does and
translating engineers Microsystems'

to overcome sensory loss in both
the auditorium balance systems.

So before I want to know the applications.

I wanted to to put up this overview
slide because I think many of us.

Maybe maybe not are a little more
comfortable in the style main.

Where we are engineering based.

But we're very intrigued
by biomedical applications.

So in the center of all that is
the biomedical Microsystems and

Neuromodulation devices that
the that our group does but

what I hope to show was
while we go back and

forth between these domains it's really
an interview process and the overall.

Goal is to overcome a sensory deficit.

So the first application
I I always go back and

forth between Koechlin vestibular but it.

Coakley or seems to make the most sense
to a lot of a lot of individuals so

I started there and these are the most
successful neural prosthesis to date.

They overcome a hearing loss profound
severe to profound hearing loss and

there's over two hundred thousand
patients implanted worldwide.

It's interesting to note that the age of
implantation is as young as six months.

And as old as ninety years.

Very very young patients
are implanted more often

these days because there's a formative
period in the first nine months.

We need to hear sound we need
to build language and if we miss

that period of the baby the brain
remaps that space to other functions.

Some individuals can do remarkably
well with Coakley or implants.

They can gain one hundred percent
speech recognition in noise.

Excuse me in quiet and noise it's very
very challenging and if you've ever gone

to a party and try to have a conversation
with someone a couple feet away from you.

It's very very difficult to even though
directionally the very close to you to

kind of mask all that surrounding
sound and for a Coakley or

implant user it's even more challenging
bilateral implants are where both

sides of someone has severe
profound hearing loss on both sides

like they have to implants and
that's becoming more and more common.

So to understand how
a coker implant works.

Let's think a little bit about
the peripheral auditory system

how it functions every day and

basically these these
transducers ahrd are so.

The ties to capture sound so
here's the outer here.

So that's designed so
we can funnel sound to.

Through the ear canal here to the ear drum
and these are an amazing set of bones

here the middle ear bones and these
are the smallest bones in the body but

they serve to both filter and
amplify sound.

So it's about a twenty D.B. game when you
go from the ear drum to the Coakley here.

The Coakley is a fluid filled can now.

So the fluid motion is what gets
translated into electrical signals.

So looking more in the Coakley of this was
that wedge that we had in the structure

and if we look at one
piece of that wedge So

here you know imagine these are filled
with fluid and these are chambers and

if we look at that structure the motion of
sound causes the fluid to vibrate back and

forth and that vibration moves this
membrane called the Bassler membrane

up and down and the balance in
memory is a beautiful structure in

terms of how sensitive is how sensitive
it is to this fluid motion and

its ability to resolve that into pit So
how does that occur.

We have hair cells here which
are mechanical to a logical transducers

we have a series of outer hair cells and
inner hair cells and

the outer hair cells serve to
their cold Coakley or amplifier so

they actually amplify the sound and there
is a feed down from the brain to them.

So there's an errant connection to
them to modulator that amplification

the actual neural conduction occurs
through the inner hair cells which

depolarize auditory nerve fibers nerve
fibers in response to this motion.

So how do we resolve pitch.

And that's through the bows remembering.

It has a variables.

Stiffness making it most sensitive to
high frequencies as you come in here.

The base the bottom part
of the Coakley up and

it's it resolves that in place
all the way up to the Apex

which is the highest frequency the highest
frequencies that we can hear twenty.

Excuse me the lowest frequencies we
can hear down to twenty hertz and

basically about in here is
where we are most sensitive.

Excuse me most sensitive to speech.

Although we have a larger range.

Those are the frequencies that
are important for Coakley or prosthesis.

So I talked a little bit about the hair
cells the transducers Now what happens.

What causes sensory neural hearing loss.

Well it would seem strange but
the one of the most prominent

reasons of sensory neural hearing loss in
individuals who have hearing and lose it.

So they're not born congenitally
deaf is damage to the the theory of

cilia that's what's depicted here and
white.

So a series of antibiotics which
are lifesaving are often prescribed

it seems strange but they're often
prescribed and they damage hair cells.

So while this loss occurs the connection

between mechanical to
electrical energy is broken.

So this can also be you can also this can
be trying to think about when it was done.

Maybe in the forty's when they had a lot
of cases of rubella mumps measles.

This can also lead to
sensory neural hearing loss.

So how do we overcome that Oakley or
prosthesis it bypasses the broken element.

So it goes straight to the neurons
the key is that the neurons

survive enough that you can activate
them with electrical stimulation.

So on the outside.

How do we make these choices about
how we stimulate auditory nerve once

I'm showing here the external part
of a Coakley or a prosthesis.

So we have to somehow capture that sound.

Make some do so that's one here on
the scale that's one hit one here

on the external portion so
microphone and they've

become much more sophisticated trying to
trying to provide directional information.

So some implants have
three microphones and

then this is where we make the choice
about what we transmit internally.

So essentially you can think of
voltages coming from the microphone.

Those are process to extract
important features for speech and

although I don't discuss a lot about
the signal processing and cocoa implants.

It's incredibly impressive.

What they've done over time in
the eighty's is when you started to see

implants in the US
population really really

revolutionary work in
speech processing with.

The signal processor here and there's
four D.S.P. course Coakley or implant.

So that information is the simplest
there's a variety of speech

processing algorithms but I pick kind of
the most commonly used to talk about here.

So that's depicted here over on the right.

So what happens is you have a sound coming

in to make those decisions about
what to send to the implant

they band pass filter it into frequency
bands that are important for speech.

So now you have this big rich signal.

You can hoarsely.

Break it up into frequency bands.

Now how do you figure out how
to simulate you know place.

We talked a little bit about that.

What about loudness level and

we do that by extracting
the energy we detect the envelope.

We make a compression just because
a dynamic range of sound is different than

dynamic range that we can stimulate
with and we apply the energy of the.

Signal.

Incoming acoustic signal
to the electrical signal so

loudness is matched to
the current level that we apply.

So looking inside.

So now we've moved to
the implanted portion

inside the body of a cochlear
implant number five.

Here is a receiver that
there is in our family.

So we receive these commands essentially.

And these commands are then used to
drive an internal stimulator and

the level is set by the signal
processing algorithm and

what's important to note here is so you
have speech coming in time we've broken it

out and now we shift that we
translate that into space into

locations along the Coakley or that we
simulate with the electrode array here.

Most select these
are discrete electrodes so

the output current into the bottom
chamber of the cult leader but

there's presumably a surviving neural
population that can respond and

there's about sixteen
to twenty two channels.

To first order sixteen
to twenty two sites.

So and then you see the neural
connection here number seven.

That takes it up so
that neural connection has to exist so

that we can perceive
the electrical signals as sound so

this isn't an older implant but
I wanted to put it in here so

we can look at some of those
dark circles mean those dark.

I guess bands the bands are electrodes and
it's in a mole did.

Silicone carrier and
to this day they're still made by hand.

It's a painstaking process these
are very expensive devices and this

is where I'm kind of trying to hint about
where we can work on some of the memes.

But I wanted to give you
an appreciation of the odd.

Quote unquote audio signal of how
a cocoa implant breaks up sound so

this is an acoustic representation for
us of how Coakley or

implant generates a signal.

So this is after it comes back
from the signal processor and

it's recontest reconstructed
into an audio signal.

Let me get back my mouse here.

So these are sentences.

I'm hearing some of you whisper.

So I think most of you were able
to maybe after the first or

second one to know how to translate
that course representation

of speech into something
that you can understand.

So that's a success story.

Now let me play some music for you.

So

I'm probably a lot older than
the students here in this isn't

quite my type of music but
I think that it's.

It's interesting.

Right.
It's what some people like but

let me tell you where it came from

so

I couldn't get it to stop.

So we've got a ways to go.

We really have a ways to go.

But these are problems that we
can approach and try to solve.

So being a meme's person.

What we started to look at is
well how can we interface to

the neurons how can we hit that
surviving population more discreetly so

what we're trying to do is
resolve more discrete frequencies

provide that richness of sound so
I'm going to talk about what we can do in

the profile auditory system but
quite honestly.

This is whatever you do
inside you always have

to link back to the signal processing.

So it really is an interplay
between both of them.

So I did what I've shown here in this
cartoon is OK here is of course activation

so hitting a broad These are supposed
to be hair cells here and the red is

what we've hit these neurons we've excited
them and the blue bumps are electrons.

So if we can place more electoral
along the Coakley tightly.

But discretely to first order we can
restrict the activation of the neurons.

So that we can resolve frequency better.

Another important aspect of this
is that the neuro population.

It changes it changes over time.

So if I was going to get an implant.

I would want it to last the course of my
life and that neural population changes.

Also the more like
TOLD YOU HAVE YOU CAN PLAY GAMES.

It's more technical than that but
you can activate and

deactivate different electrodes that
are close so that you can do some.

Current steering and current shaping
that's a more power hungry strategy but

it's been shown that some of these
techniques are very important for

those individuals who have a very
restricted body neural population.

But the challenge is we can't take
existing electrodes that are made

there in case silicone wires.

If we try to put more electrodes that
bundle just gets fatter and fatter.

So then what did we think about what
we thought about memes technology.

This is the University of Michigan
work that I did for my Ph D.

but I wanted to highlight that
I did the electrode reports and

we were part of a large center so
we could make a complete marker system.

Which the chip here has its own button.

So but
what I wanted to point out here is that we

just I think I'm not going to let
myself touch too many things.

But I wanted to point out here
is that this has a D.S.P.

core as well as a microcontroller
that we developed but

where I spent a lot of my
effort is validating since film

silicon thin film array
in an animal model.

So what.

How do we make these devices.

Well the.

Kind of the fun thing about the Michigan
technology and silicon is that we can

adjust the profile meaning that
the depth of war on doping is that stop.

So you can have a bigger portion and

I don't I don't know if any of you have
tried to water to anything you really need

a robust substrate to wire bond to
those Gold Bond pads back there but

going into the Coakley if you
remember it's a spiral structure.

It's very very challenging to put a thin
film device in that spiralled structure.

So the thinner you can
make it is three microns

in depth the thinner you can
make it the better off you are.

What we did is we we made
the electrodes very small and

very close together and this was
because we used a lithographic process.

So I won't dwell too much on the.

Fabrication process but it's pretty
straightforward in terms of using the born

as an edge start layering dielectrics
putting down metal stack.

Larry more dielectrics and

the radio oxide stimulating sites
which is a little different.

Most electrode arrays cochlear
electrode arrays use platinum

platinum is very stable and
then a final Senate.

You do an E.D.P.
etch which is a pretty aggressive.

Either way the silicon process.

And our electrode So
I wanted to point out here that we have.

A commercial.

So this is a half banded or
a this is a more contemporary array.

So the electrodes are in half bands.

It's pretty quiet.

But our electrode sites were
a tenth in terms of area and

so if you think about neural
stimulation you really

think about you need to get enough charge
out there to depolarize the nerve fiber so

when you scale down an electrode site or
surface area.

You really have to get smart.

About the material that
does the reversible

redux reactions that the surface to
ultimately depolarize the neuron so

that's why we went to a radium oxide
what I've shown here are thresholds.

And basically we wanted to show
that the threshold means how

much current does it take to get a neural
response where I was sure that we've

switched the technology just something
along the lines of thin films but

the amount of current we need
is similar to contemporary So

most contemporary arrays the stimulators
are designed to go from ten micrograms

up to about one to two million
amps which is really really high.

So we were in the range that
we would expect to see.

So moving further great
Silicon is wonderful.

You can do a lot with
silicon is just not good for

the colder application because although
it's flexible the Coakley a spirals and

if you put in an electrode array it
actually needs to twist to go from

the base of the lower the higher
frequencies up to the lower frequencies.

So when I moved here I started working
with some folks at Georgia Regents

university who were really very clinical
but they were really motivated to

look at the technology and I said well
we're going to have to go to a palm or

we're just going to have to go to
a polymer to get these devices in.

So what we did is we used
a thin film array that the T.F.

a it's made out of poly in it.

And this is something I wanted to
do during my Ph D. I said Well.

Can't we just use commercial arrays and
put our thin films on them.

So this what I've shown here this.

These are industry names for
silicone based.

Electro to raise their nonfunctional and

I T V is not really used by a surgeon but
it's there if they somehow want to.

Insert it to check the path for
the electrode array and

that can cause damage itself so

most surgeons don't use the I T D I even
an insertion electrode it's just a blank.

It's just pure silicone so

we were able to work with a clear
implant company medal out in Australia.

To let us try to test out our devices so
we adhere to them kind of a quick and

dirty approach.

So let me point out a few things here with
there's a cross section of the Coakley

are and what this is they are their

insertion test device and
this is our thin film array.

It was not the greatest to do because
it's very very hard to get here

something to a dissimilar material which
we kind of should have guessed it but

I'm pretty stubborn so I wanted to try it
and so these were done in human cadaver.

Coakley up so
it was not really really good but

we did learn that we can at least the
surgeons got comfortable working with our

electrodes so we took it a step further.

What I will go through
every detail here but

I want to point out this
probably this picture.

I think is the best the bottom one to
see you can see the electrode sites.

So we came up with the way
to precursor the device so

we thought well the Coakley is curled
let's give the surgeons a pretty

curl device so we'd hear the OR A It was
a more robust strategy we coated it was

silicone it turns out that the surgeons

didn't like our race because they're
used to working with the straight or a.

So we had a new resident and

we actually didn't we only
implanted fifty percent.

We were only able to implant fifty
percent of the time in the Coakley.

So that was that was interesting for
us as well.

So.

What we decided to do is say OK fine.

You don't want to call it.

You don't want to bet.

Let's just coat them a little bit
because since the morays I have

some here we can look at him after a
really hard to handle extremely difficult.

So we said let's just put some silicone on
and we'll give you some points that you

can touch that don't damage
the are a very simple process here so

we coated them using the same type of
equipment that applies solder under

pressure solder paste under pressure and
then we put these in and

animal model saw a live press in cat.

And stimulated and looked at the response.

So the results were actually pretty good.

I wanted to take one second here and

talk a little bit about how we do these
experiments one slide because I know this

is along the lines of electrophysiology So
how do you do a perceptual experiment.

So what we do is we input a stimulus and
we record electrically So

there are specific structures in
the auditory system where you can

record one of the easiest is
the inferior quality lists.

So we're going along the path
from the per free to the century.

So we're going from ear
up into the brain and

there's different points you can probe for
our tests we actually this last

set of experiments we did in
the cat we probed externally.

We looked at an external response
that they do clinically So

the next step is OK there are cells
other transducers come in apply

an acoustic input and record make sure of
the acoustically that structure is sound.

Then we apply the series of antibiotics
that I told you cause here so

loss and in the time course of a half an
hour we've induced deafness in the animal

model not the greatest thing but
that's how we do the experiment.

So once we validate that we have
definitely cats that we keep applying

it to seek input and we get no response.

And we know we've succeeded.

We come in and employ apply.

I've shown a bi phase a pulse here so
essentially we balance the charge going

into the system and
then we look at the response.

Here's a characteristic response of
an auditory brainstem response that can

be measured externally the first portion
here is stimulus artifact because you're

you're you're putting in a lot of
current and so the recording system

responds to that the current itself
not the auditorium response so you.

Ignore that and then what you
see the kind of jumps back and

forth here are analyzed by
electrophysiologist to be

indicative of an appropriate response
to electrical stimulation and

when we do these experiments we want
to know how much current does it take

to get the response so we look for that
threshold and that gives us a sense of

you know if it's way out of bounds we
know that we may have a mis placement or

it's a funny electrode or
if we can also look for

open circuits possibly when
we inserted the we broke it.

I mean these are kind of the real
world experiments that we do so

we got a series of responses
that were appropriate.

You know one hundred seventy micrograms
is a really really decent number for

current stimulation.

So the next thing is looking at where the
electrode array is in the Coakley Let's

see this one looks pretty dark Let's
see if we can get this to work here.

So this is a micro C.T.

image and we have access to really really
great imaging machines at Emory and

so we can have twenty Micron
resolution this is not for a human.

So this is an animal.

Get it back.

This is an animal model here so
that machine.

Go there.

You should have my phone
in the right spot so

let me play it rather than
keep talking through it.

We're coming up through the Coakley.

Did you see the electrode array.

So these are nice slices that give
us a sense of where did we put

the electrode array.

Especially when you have an experiment
where you see some facial nerve

stimulation you can would happen one three
see the cat their facial nervous twitching

so you know that you're over stimulating
and then this particular animal model

we harvested the Coakley and
we imaged it to see did weaving in and

did we get it in the coco we thought
we did in this case we did so

this is just for the electrode array
is you can see the sights hopefully.

And so this is really significant we
weren't able to do this years ago.

So it's another level of
validation of where the electrode

over a is in the Coakley.

Because you need to be close to
that surviving neural population.

I think in this one just shows that
we measured it appropriately but

let me pause here before I
launch into the vestibular work.

Are there or are there any questions
about some of the Coakley or

electro work in neural stimulation.

So the status of this work is in terms of
chronic really what you want to get to our

long term chronic recordings there are
some challenges because polymers are good.

But they're not as good as medical grade
silicone and platinum radium wires.

So we need to look at chronic recordings
as well as I talked about high density.

So all we've shown is that we stimulate we
get a response but we haven't been able

to do enough experiments to look at
if you step through electrode site.

If you compare a commercial to our ray.

Are you really providing a better
resolution and to do that we have to

measure in the unfair or click us not
the course auditory brainstem response.

So that's in this next set
of work in the still main.

So moving to the vestibular Biosystems
they're very similar to Coakley or

prosthesis the system level but so
what's the what's the challenge here.

First let me point out
the vestibular system.

This is is located here in the inner ear.

So we have three semi-circular canals and

the sense angular head rotations leaving
my head like this and I'm activating my

semi-circular mostly the horizontal
semicircular canal and there are three of

them nearly mutually orthogonally designed
so we can sense three D. angular rotation.

Also in the otoliths which
are located in the vestibule here

we're sensitive to gravity and
linear motion.

And so I when I started this work.

I really didn't know much about
the vestibular system but it seemed like

a logical extension of the Coakley or work
and I had to put together a chalk talk for

a faculty position somewhere OK will
do will do just to be a lawyer and.

It turns out that this system
is very very interesting because

it's responsible for maintaining gaze.

So when you move your head is so
imagine you're sitting in a chair and

it's rotating but
you need to fix your gaze on a person so

your eyes have to move and
I keep looking at poignancy so.

Your eyes have to move as your head moves
how do we control that it's the vestibular

system more perfectly it's
controlled through a reflex but

there is the most serious is a very
challenging because there's layers and

layers so we look we want to know
how we how we are moving in space.

So it depends on what our muscles
are doing what our head is doing so

it's a highly integrated system but
a failure in the periphery can lead.

To vestibular dysfunction.

So let me talk a little bit
of vestibular dysfunction and

I will because I've learned more and
more about this as I work with the group

over at Emory and now they're in
midtown just down the street and.

Vestibular or dizziness itself so

not all dizziness is associated
with vestibular dysfunction.

But over forty percent of us will go to
a clinician because we have this in this.

So that's pretty significant also in the
elderly population and sixty five doesn't

seem that elderly anymore and nearly
nine percent of these patients will go.

Because they have a problem of balance.

So balance is especially
dangerous I would say in

the elderly population because
of falls and falls often.

Well I mean the slide says it all
fatal and non-fatal injuries for

that population.

So it's pretty it's very significant.

So let's talk a little
bit more about that.

Well what's the good news and the good
news is vestibular rehabilitation so

that means training.

We can overcome a lot because
we have a nice duality we have

to vestibular peripheral vestibular
sense systems you could say.

And if we lose one we can
compensate with the other.

Those who lose both they don't
do as well but we still try.

Now the bad news is for those who have
severe bilateral vestibular dysfunction.

We can't compensate with fear.

And that's one of us to deliver
prosthesis comes into the domain.

But I wanted to talk about the good news
first because I wanted to look at so

we're talking about memes devices but
what about using existing memes devices.

And patients those mostly with
unilateral vestibular dysfunction.

We can improve their gaze
stabilisation which really helps.

So some patients can't they cannot
even go to Wal-Mart because

as they're walking all
the you know rows and things

just make them very busy very very dizzy
so we have a set where they have a set.

I don't I observe I don't prescribe
anything if you do these gay

stabilisation exercises which are really
hard to do you do them at about one to two

Hertz of head rotation rate two to
three times three to four times a day.

One minute each and

you're looking at a card either in front
of you or pasted on a wall ten feet away.

And you're trying to get
your brain to compensate for

your You're basically trying to lose
the game on your vestibular ocular reflex.

That's the reflex that senses here and
moves these so

we've shown that they've shown that
seventy five percent of these patients can

improve twenty five percent of
their of them are not improving.

Why are they not improving they feel
that one the exercises are hard to do.

So those subjects may or
patients objects may not be doing

them at the appropriate rate at
the appropriate durations So

what we did is we took a very simple
system and we embedded it in a hat.

So this X. ray we took an X.

ray image of the hat so
we could look at all the pieces.

This is a nice try.

Axel gyro from invent sense and
we just you know we got it on a spark on

board this is one of those projects which
is really a lot of fun very inexpensive.

So we capture head motions we process
it with a microcontroller and

we can then save that data.

We also have a wireless interface for
this but

I didn't show that we can look at
a patient's profile really quickly.

Over the exercises you can get you know

the amount of time they're doing
their exercises very quickly.

You can look at are they doing them
better to the right or to the left.

Are they fairly regular.

So these things which
are really simple to us.

I mean this just look like squiggles
of the Matlab gooey I mean for

engineers is fun but for them.

This is the first time
the physical therapist and

their all ages are able to have this data.

Readily available.

So that's why I call this
the good news portion.

So now let's move into the prosthesis
work which is really quite challenging.

So Dan Merrifield shot his
name down here at Harvard.

He was the first to develop
a vestibular prosthesis.

And I have to say the reason why I got
into this work is Dan came to me on

neural prosthesis workshop
in two thousand and three.

Two thousand and two and he said
Can I borrow one of your Coakley or

electrode arrays to use
in the a similar system.

I want to try to use it to stimulate
Mr Miller neurons and I said really for

what it's of well we're
building this prosthesis and

we'd like to try using your race.

So what does it do well to first you
know the first easiest prosthesis

is to hook up a bunch
one to three gyros So

this is his work was in the horizontal
plane first simplest to approach.

So angular velocity sensor gyro signal
conditioning aided the conversion things.

The police love.

You go ahead in their map that into
a signal that's appropriate for

a vestibular neuron stimulation and
it's rate based.

So again we have that characteristic nice
by phase in charge balance polls but

instead of modulating place as
we do in the Coke we are we.

Module A rate.

In the vestibular system and they showed
in animal models that they were able to

get stabilise gaze in response by
activating the vestibular system.

So.

You know we did some couldn't get a good
to go back and forth a little bit but

I wanted to just point out
these structures because

they have a meme's I guess analog so
in the semicircular canal there's Is

that a friendly like thing is kind of
a weird gelatinous structure called

the the Ampere law is a pink thing and
in it.

The Green Thing is a coupe ULA and what
it does is these are also fluid filled.

So if we move our head.

We're activating the vestibular
system the fluid moves it pushes

the green the ampere and
we have these sticks in here.

These are again hair cells and
their deflection sensors so

we map this deflection
into a pulse rate and

in the vestibular system we
activate it with acceleration.

But the pulse rate codes
instantaneous angular velocity.

So the semi-circular canals themselves.

Do one level of integration
on to that signal.

So there's a new one more level of
integration that's done above because we

move our head in the opposite direction

we move our eyes in the opposite
direction to our head.

So there are some pretty interesting
signal processing that goes on within our

system.

We also have a baseline firing
rate if you're not moving.

They're still firing.

And as I move this direction this
firing goes up and this went down.

So in our brain.

We're actually taking signals from
both sides and using that and

some of what rehabilitation does
is when we only have one side.

It makes us use that side.

So we take the richness of that signal and
use it as best we can.

So how do we test
the validity of stimulation.

Is we look at the motion of our eyes.

So I wanted to point that out.

They do this.
Routinely

it's rotary chair testing it's
actually not very fun but

you sit in a chair and they rotate you and
they look at the response of your eyes and

you know it's a basic system
identification task because they do this

over different frequencies and
they get a frequency response out of you.

And so they look at your gain.

As well as your face and they can diagnose
of a stimulator deficit that way.

Yeah that's me.

Those are Miles.

It was no fun and
they told me I'm a little bit.

I have a little bit of imbalance
in my wrist and it was just so.

I think clinical So
now we got a little exciting and

we thought well let's see if we can
model that gelatinous structure why

don't we try to make our own
vestibular semicircular canal.

So this work in terms of
understanding of US A real or

system as a mechanical system was
done in the thirty's and forty's.

So they came up with the second order
model what what's depicted here.

This is the semi-circular can
now we've made it a Taurus blue

is liquid liquid in our summer circular
canals is very similar to water.

So it's a nurse will element so

as we rotate the liquid fluid moves
against the coup people bends the coop.

And that's how the that's how we send the
signal the fluid has a viscosity that has

in us the city so to first order we can
model our vestibular system by a spring.

And a spring in that kind of system here.

So we tried.

We're trying to make a meme structure
out of that it's kind of fun.

So this is the early work in modeling
that in consol which is no no easy feat.

As we can see here we've
spent a lot of time on that.

So our first model is well OK let's me.

A sensor we needed die for
and we need the coop.

OK memes people are great at making
microphones and things like that so.

Can't we try to make a memes
type you know diaphragm.

But then we need senses so
we need a reference electrode So

it's capacitive sensing and
basically the diaphragm moves and

that is what codes the angular
acceleration or velocity so

you have to adjust the rest of the
structure to give you that integration.

So some of the early work that.

Our group did here was on.

You know looking at the frequency
response of Kugler displacement and so

that one of the other challenging pieces
is not just in the marker fabrication but

we have a Y. dynamic range in our body but

in member structures we're always trying
to linearize if we can linearize and

we can premier tries and understand but
we have a huge dynamic range so

quite obviously one one membrane
is not going to do it for you.

I think we need a series of membranes and
so

we may have some early prototypes and
I won't dwell too much on this but

this is again the semicircular canal and
our natural come out.

We do have that ambulatory organ so we do
get wider to make us more sensitive in

that domain and we made a sense and
reference electrode these these were

larger scale devices what we use
pm M A S U eight micro mold so

kind of a quick turnaround and
sputtered gold for the sense and

a reference electrode so
that poses some challenges.

I don't think we have
any breakthroughs there.

And I wanted to launch into case a let's
go back to the vestibular prosthesis I

wanted to bring us back
out to the system level.

And what work.

Have we done there and it's a lot of
that work is on signal processing.

So it's not just on the mems devices but
it's looking at how we.

Be smarter about coding angular velocity.

And generating by facing.

Says so contrarian who is here and

he's a pretty soon to graduation
has done a lot of this work here.

And what I wanted to
show is how he has come

up with a strategy to do the signal
processing very smartly.

But what is needed.

Well we don't just need the code angular
velocity and generate current pulses.

We also can provide a better
therapeutic benefit if we can then

use information from all three canals and

provide a correction so let me back up for
a second number one when these.

Gyroscopes are implanted into a patient
the time in the O.R. is expensive.

So there's not a perfect mapping of
where they implant the gyroscope

with the canal plane.

So the doctors I work with say well
it's better if you would just go and

put it in and
then we can correct a little bit later.

So the other pieces now are empty
Larry nerves the nerves that we want

to stimulate are very close and so what is
shown here is here's a horizontal canal.

Here's a POS theory or canal where we come
in with an electrode and we're just trying

to get the horizontal canal but
guess what we get the past year ear canal.

So if we can provide a pretty compensation
for that signal we can provide a more

efficacious therapeutic benefit to
the subject because if you come in and

you get implanted in there and now you're
more dizzy because your other canals

are being stimulated I don't think that's
very that's not where we want to go.

So I'm going to have to hop
a little bit through his stuff but

I'm sure he'd be happy to tell you more
about it is that he does low power

analog signal processing by
operating his circuitry his

circuitry in some threshold and what he
did is time division multiplex so he's to.

Three signals all three canals
applying a correction by mapping it's

a vector matrix multiple multiplication
weighting the weighting the each of

the canals differently and
he's generating an overall response so

that he can correct for any misalignment
or patient specific parameters here.

So I'm going to hop through this just in
the interest of time but a nice a nice

power number less than ten microwatts for
this chip that he developed and Ts MC

point three five micron process and he's
also has a T.I.A. chip that he's testing.

So this is a lot of fun.

I want to close with the number of
students who have participated in their

research and you can see that we in our
lab we have a very diverse student group

because we also work with residents and
medical students because we go back and

forth between the different domains of
medicine and engineering and I also want

to point out that we have a very strong
strong set a very good undergraduate

students because some of these projects
that take existing men's technology

undergrads can really do well and get you
know hit the ground running with those.

So in closing I wanted to acknowledge
our funding sources for this work.

Our national funding sources as well
as our international partner meadow

in providing devices a lot
of support to us and

the last slide is really the first
slide and I that you've gotten some

ideas about maybe where your research
may go as we integrate the engineering

with the medical or
diagnosing applications and I didn't

talk much about nano mechanical sensing so
that her cells are now mechanical sensors.

But maybe you have some ideas about how
you can bring these domains together.

Thank you Paul.

It carries the magnitude but
the frequency is place.

So the very that's a really good question
for low frequencies we can track.

So the very first implants were
single electron single channel and

people could gain or only a couple they
could gain stay could gain some sensation.

And there's always there's always work
that revisits that concept can you get

away with but for
the higher frequencies you.

It doesn't there's not a one to one.

It doesn't it doesn't I mean
the membrane is more sensitive

mechanically more sensitive
to specific frequencies.

So we are doing.

We are doing signal
processing mechanically but

that is an important point
because we are making we are sort

of saying that a place is frequency and
that's not necessarily true for

the low frequencies you can provide
rate and people will understand pitch.

There and what was actually interesting
is that the pulse rate I didn't put

up the pulse rate but it's eight
hundred thirty three pulses per second.

So that it is amazing how
we can sort that out and

understand that perceptually but
it is a question to keep revisiting

thanks.