Non invasive deep brain stimulation via delivery of temporally interfering electric fields


good afternoon everyone it’s a pleasure to be here I hope you all hear me well there are billion people
around the world that suffer from brain disorders costing our world economy
around trillion dollars every year brain stimulation technologies that affect
activity of the brain can help millions of them so the overarching aim of this
project has been to bridge the gap between – brain stimulation modalities
so one side we have invasive brain stimulation you can see here on the left
where we in insert electrode inside the brain to affect activity of small deep
structures in a very focal way on the other hand what you see on the right we
have a non-invasive brain stimulation modalities that using electrical
magnetic fields to affect activity of the most superficial structure of the
brain called the cortex however in a dispersed way specifically what we aim
to develop is a method to target in a 3-dimensional vocality deep brain
structure in a steerable way that means we can move it around without moving the
electrodes and to do that without chemical or genetic manipulation of the
brain tissue so in this presentation I’ll show you I introduced you this new
brain stimulation modality that we developed and take you through a series
of tests that we have done to validate it first in an animal model and then in
humans I will go pretty quick through some of
the data store ever if you have fundamental questions about concept that
I’m using please stop me and raise your hand okay so the new brain stimulation
technology is braised on the physical principle of temporal interference in
temper interference we are using we applying one current
imagine this is now we’re presenting our head
so in tempering the furnace we apply one current via one pair of electrodes on
here in blue in black at the frequency that is a killer frequency that is too
fast for the brain to follow at the same time we apply another current to the
brain so any another pair of electrodes showing here in blue at the kilohertz
frequency that is slightly different from the first one so for example if the
first one was 1000 Hertz the second one could be from stampin 1,010 Hertz these
two currents create electric fields inside the brain shown here in in black
and blue arrows this electric fields are superimposed by the neurons inside the
brain if we look at it on the time domain each one of these fields shown
here in black and blue oscillates change periodically at the frequency of the
head however the superposition field shown here in red has an amplitude that
change periodically due to the superposition at the frequency that is
equal to the different frequency of the two original currents so for example if
we apply 1,000 Hertz 1,010 Hertz this amplitude will change at 10 Hertz which
is the different frequency the amplitude of this change of this is proportional
not just to the amplitude of the original currents but also to the
relative amplitude and relative orientation so for example if we’re
looking now in this location inside the brain deep inside if we look on the
field just zoom near we can see that the two field have similar amplitude and
similar orientation and therefore we have a relatively large amplitude of
this change however if we look closer to one of the electrode pair when one of
the field is significantly smaller than the other and has an orientation that
perhaps perpendicular to the first one then the amplitude of
this envelope envelope change in the amplitude is magnificus molar so to be
able to test that we have to this concept of this perished concept we have
to develop some electronics to allow us to inject two currents inside the head
in an independent way they are sharing our head is a is conductive so if we
apply let’s say to plus and minus there’s nothing to stop one plus to go
to the two – creating crosstalk but what we want is to be able to precisely
control the pathway of this current at applying into the brain simultaneously
but in the independent way I won’t go into details but we develop a technology
it allows you to do exactly that it’s a different way by which we inject current
what you can see here so here will be one channel of applying current okay
that’ll be kind of a schematic of this circuit that’s a protein and that would
be another circuit at applying a second kind between these two electrodes this
will be frequency number one and it’s the f2 if we are measuring the actual
contact and these two points shown here in a normal case we will see a
contribution from the f1 which is the frequency we inject here but also a
large amount of current that is coming from this pair of electrodes showing
here in bloom that means we will have this what we call crosstalk that occurs
near the electrode and the two will not interfere or the peak of the
interference of superposition that will not occur inside but the electrode so
with it with the method that we develop what we call anti Farrakhan drive we can
eliminate that and this each one of these points we can control precisely
what we inject to allow to separate them and now the superposition of them
essentially inside the head so the first thing we wanted to test is if weak if
the envelope modulation of this temporal interference can drive action potential
activity in the live brain X potential is essentially the way by
which neurons communicate in with each other and we wanted to see if we can he
can induce that generate that inside the live brain so to do that what we did we
put electrodes on the skull of a live mouse and here it is the two of them
each one of them was pair with another electrode that was sitting on the thorax
of the animal and we measure the activity of single neurons during in a
large brain using a technique called patch clamp which is essentially coming
with them one millimeter one micrometer diameter size electrode all the way into
a single neuron into the light brain and measuring its activity what you can see
here that when we have when we apply 10 Hertz of stimulation which is a normal
frequency by which neurons operate inside our brain we can see that we can
create generate activity and that 10 Hertz so if you see here zoom view that
would be an action potential so that we can generate in controlled activity of
this single neuron in the live brain we did a normal in a relatively reliable
way in with the low frequency or normal stimulation however when we change the
frequency now from 10 Hertz to the 2 kilohertz what we can see that after so
transient there is no activity that is generating or don’t use exactly the same
electrode and exactly the same current amplitude but now if instead so if we
use the same amount of current so the current here is 200 micron but instead
of applying 2,000 Hertz we applying 2,000 Hertz in 2010 Hertz what we can
see here when the two currents ramp up sufficiently the neuron inside the brain
start to generate activity at the 10 Hertz which is equal to the different
frequency in the same way as you will do with in low frequency stimulation what
we also investigated to see that the transient activity that we have seen
here how it depends or half we romp the current what you can see
here is what we had before when we romp the kind relatively short with zero with
25 seconds and we can see that there is kind of a transient effect happening
here when we just applied 200 or 2000 Hertz or two kilohertz however if we do
that now at 0.5 seconds so it’s lowering the ramping period we can see that this
is a transient effect does not exist and we measure that both in the cortex but
also in a more deeper structure called the hippocampus and we validated the
ability to drive action potential activity again in different structures
in the cortex and the hippocampus so what you can see here this is the firing
rate so when we apply the stimulation what is the frequency by which the
neurons operate so with applying lot Convention and 10 Hertz the neurons of
spiking on generate action potential and 10 Hertz if we do 1000 Hertz on 1010
there’s no difference between them again generating 10 Hertz of neural activity
in the same way if we do 2000 and 2010 Hertz but if we apply the kilohertz
frequencies without the difference between them for example 1000 Hertz or 1
2000 Hertz there’s no difference between unless activity of the neurons oops
between the what we measure and the spontaneous pairing and the same is true
what we recorded in the hippocampus any questions about that
great so after we validated that we can generate action potential activity in
the live brain we wanted to better understand how the amplitude of this
envelope change in fields depends on the Gamma tree of direction remember we have
two pairs of electrodes so to do that we simulated and measure the electric
fields in a phantom in this case is simply a petri dish of 50 millimeter
diameter then we put a conductor on inside
so what you can see here this is these are the result of the simulation and
these are the measurements these are the field on the envelope modulation at the
X this is the X direction and this is what you can see here on the right
direction so what we found that if we put the electrode at are preserved gamma
tree so we have reply time between a grey electrode pair and another con
between the black electrode pair the trapezoid gamma we found in this
configuration the envelope modulation is localized at bit at the center of the
narrow base of the trapezoid as you can see here but and close to the surface of
the of the phantom however if we increase the size of the narrow base of
the trapezoid as you can see here keeping the wider base of the trapezoid
fix the location of the energy undulation is is shifting toward the
center of the phantom and if we have a kind of rectangular configuration so one
current is applying it this way and another kind is applying it that way the
location of the N word modulation is shifted to the center of this phantom
however with the trade of that the area that been covered by the envelope
modulation is larger by a factor maybe of three and the amplitude that
generated tip inside is a factor of eight to ten lower
so if here zero point three eight three point eight meaning there is a trade-off
of how much vocality we can achieve a depth so to test whether we can use this
principle to recruit deep brain structures subcortical structure without
the overline cortex we put electrodes on the skull of an animal of the mouse
above the area corresponding to the poe compost program which is important
structure associated memory and we stimulate it for about 20 minutes 10
seconds on 10 seconds of and then after about 90 minutes that we let admin
rest we allow some early gene expression that are proportional known to be
responsive for neural activity and specifically there is one called cephas
that is commonly used to measure change in activity in neurons so we stand for
that at that point so what you can see here this is an example of you can see
there so this is a 10 Hertz stimulation so this is correspond to where we’ll be
one electron this is a slice representative slice so here will be the
Pugh campus this would be the cortex and this would be another coating this is
the other hippocampus so these represent where the electrode would be someone
actually is here the second uh near each one of them is paired to a second
electrode is sitting on the thorax and what you can see if we look at the zoom
view and I hope you can see it from there there is a large you see this kind
of a bright points indicating there is an expression of this C force that we
stand for meaning the neurons were activated and
this area underneath the electrode as one will expect but the fields were
strong enough to reach all the way to the hippocampus and you can see a large
activation also in the hippocampus when we did exactly the same but now with the
different frequency the frequency was 2,000 Hertz and 2000 Hertz you can see
here 2 kilohertz and 2 kilohertz we did not see any expression in the cortex
underneath the electrode the correspond to this point and the denotes the
expression in the hippocampus but now if use exactly the same current and but
change one frequency from 2000 to 2010 so having a different frequency of 10
Hertz we could see that there was a significant activation of depot campus
as you can see here in a similar way it was done with the conventional
stimulation however within contrast to conventional stimulation there was no
activation of the cortex above or overlaying the hippocampus indicating
that we were able to recruit essentially a deep brain structure
in a non-invasive way and without the overlying cortex and we obviously
validated in a population of animal and I won’t go to bitters but we can see a
significant activation of the paw campus when we compare it to the contralateral
size the Snyder was not stimulated in the same way it was done with
conventional stimulation but no activation of the cortex here while in a
normal stimulation obviously there is a significant activation underneath the
electrodes which made the cortex so during the physics experiment when we
measure the electric fields we notice that the location of the Animas
modulation depends not just on the geometry of the electrode but also of on
the current ratio the ratio of these two currents so before I show you an example
that we had two currents so we have this kind of rectangular configuration so one
kind is applying like that and one another con is applying like that with a
similar amount of current so let’s say we apply one milliamp and here and
another milliampere the total sum is 2 million and the Russia between the crown
is 1 to 1 but now if we decrease the current on the grey electrode pair and
increase the current on the black electrode by the same amount
keeping the current some fix to 2 million we can see that the location of
the annual modulation is shifted or steered to vote the electrode pair with
the lower amplitude and at the distance Babbitt is steered is proportional to
the ratio of discount so here we had a ratio of 1 to 2 point 5 here’s Russia 1
to 4 as you can see it’s dear father so we wanted to test if we can use this
principle essentially to functionally map a brain region without physically
moving into electrons in order to do that we put one electrodes or above the
cortex motor cortex area that corresponding to the movement of the
contralateral PO so here’s one electrodes one electrode correspond to
the movement of the contralateral palm and we put another
roads on the above the other hemisphere but the area that corresponding to the
whisker movement of the electrodes it’s the Letran on the same side as was
apply here so when we if in the motor cortex the right motor quarter
correspond to min the movement of the left side and the left motor cortex was
point to the movement of the right side and so on so what we did we stimulated
that and different current ratio and measure activities of the pose and of
the whisker so what you can see here when current when the current ratio of
i1 and i2 is 1/2 a that means that this current here is significantly lower than
current i2 predicting from our modeling that the location of the emblem that
will be still stored at so steer toward this direction and therefore potentially
will activate more designs and indeed what we can see we see a large
activation large movement of the controller of the poor shown here in
blue and it’s an large activation of the whisker and as we decrease or as we
increase the current of my 1 and increase the kind of i2 by a semi month
so you can see here the movement of the contralateral toe this one decreases and
the movement of the whisker increases social indicating that we are steering
our stimulation from this point molted at that location but now and this was
done without again moving the electrode so now if we are further increasing
current i1 to be larger than item for example as you can see here we start to
see movement of predicting that the stimulation we start to steer to this
direction we start to see activation of movement of this poem which is epsilon
try to i1 and however non significant but the significant activation of the
whisker as you can see here in green with the arm P that increase as we
increase the amplitude of of i1 I won’t go to details but we also
characterize how they threshold so how much current arms we need it depends on
the frequency that we use we didn’t see a big difference between trying to evoke
movement at 1 Hertz or different frequency of 1 Hertz a different
frequency of five Hertz at 10 Hertz and so on
however we we found that in comparison to conventional stimulation which is
will be in this case 10 Hertz we need about 3 times larger current to
evoke movement with with thousand and thousand and ten Hertz and as we
increase the carrier frequency her frequency is the key lowest frequency
since here we had one thousand one thousand ten he is two thousand thousand
and ten we need the more and more current almost in a linear relationship so to start testing if we can use this
new brain stimulation modality in humans if it’s safe and can recruit activity in
the human brain we stimulated the motor cortex of of healthy volunteers
targeting again the motor cortex so we had a similar trapezoid configuration as
you can see here so one current is applying like that and the second kind
is applying like that and we recorded concurrently MRI
activity of functional MRI activity which is a way for us to measure
indirectly change in neural activity by change in a relatively oxygenated blood
activity and we tested different control stimulation so we have a normal 10 Hertz
stimulation we had that 2 kilohertz stimulation and we had Chomp which is
meaning everything was the same but the current amplitude was set to zero and we
had a temporal interference stimulation with one to one current ratio so for
example one milliamp and 1 million and we had the same temper interference with
three to one current ratio and one two three on Congress shoe examining or
testing the kind of steerability of this principle
and we repeated each one of them for time in around the modem to be able to
characterize the effect on the brain what we did we to use the structure MRI
data from the subject you see an example and created a model of that pinnate
element model that allow us to essentially simulate where would be the
distribution of the envelope modulation or the fields and use that essentially
to say where is the peak where is them so where is the maximum and we treasured
that and one over in which is a fraction of this maximum and said this is the
area that they mostly exposed this Android modulation and we wanted to see
what’s happened there so we define that as a region of interest this is what we
want to do because the brain this was a rest condition so the brain does many
many things in Parlin but what we were interested is what exactly the
stimulation did and we define we accept we split the region of interest to a
frontal area and a dorsal area and measure basically the activity or the
function of what we called ball signal or function M R activity them so here’s
an example of of one subject so here we applying 2010 and 2001 of them has 1
milli amp of current and when we simulate this this specific subject
fields we can see that so here’s one kind of line like that and the second
planet that we can see that we predicted and Vevey ssin will be somewhere in the
center and this is the this is the actual significant cluster that coming
out from the functional MRI of this particular subject and you can see here
so if we zoom here and these are our frontal region of interest and this is
our dorsal region areas and in yellow we see the kind of significant activation
the seed activation significally happens at the center between the two region of
interest and when we look at the core group level we don’t see a significant
difference between the frontal side and the dorsal side of of the region of
interest but now we decrease the current of the
frontal electrodes this one from 1 million to 0.5 million and increase the
current and the dorsal electrode by the same amount so we keep the current sum
to 2 million predicting from a simulation that the activation on the
anvil moderation will be steered frontally and indeed if you look at this
subject specifically you can see that the cluster of activation move frontally
and the subject and at the group level we can see a large activation at the
frontal region of interest compared to the dozerman and now if you do exactly
the opposite we increase the current i1 or the frontal 1 to 1.5 and decrease the
dorsal / account in a dorsal electrode pair so again keeping the current some
fix to 2 million predicting from our simulation that the amber medulla should
be steered or saline we can see in this subject that the cluster of activation
move dorsally and on the group level we can see there’s a large activation of
the dorsal the backside of the region of interest compared to the frontal mark
indicating that a we were able to recruit activity in the cortex of humans
and we able to steer the location of activation simply by changing the
current ratio so to summarize temporal interference is
a new brain stimulation modality that is based on the difference of multiple
kilohertz frequencies the physical principle of temporal interference has
been known for hundreds of years and in fact has been used to recruit
neuromuscular structure we further developed this physical principle and
make it relevant to the brain we show that we can use this principle to drive
action potential activity in the light brain we can recruit subcortical
structural deep structure without the overlying cortex and we can functionally
map a brain region without physically moving the electrode and and we tested
started tested it in human the temper interfering
does not yeah question correct no so that’s it so the typical pathway
of new technology will be done first on healthy individuals and then we take it
into patient population there are obviously challenges in in
many I spoke of doing it part of that is they need to really understand how the
neural activity is underpinning the disorder that we are trying to fix some
of them are obvious some of them an or unknown someone did not okay
just to finish that so the so the temper interference stimulation does not
replace existing brain stimulation modalities such as deep brain
stimulation buying implanting electrodes or non-invasive transcranial stimulation
by putting superficial electrode by instead adding more capabilities to this
ecosystem we hope that we do the stimulation without chemical or genetic
manipulation of the brain it will be used in the clinical research were
eventually Richard therapy as you mentioned there are many people that
contributed to this work that was partially done will be done in MIT in
Harvard I particularly want to mention my mentor there at Boyden and all the
different funding industries specifically and then UK dementia
research in thank you for your attention you

Leave a Reply

Your email address will not be published. Required fields are marked *