Transcript How Do We Do It? (2)

Nerve Chips: Bridging
Mind and Machine
Alik Widge
MEMS Laboratory
Neurobotics Laboratory
Carnegie Mellon University
Your Humble Speaker


Dartmouth Class of 1999
 Double major, computer science/cognitive science
 Inspired by ENGS007, Fall 1995
M.D./Ph.D. Program, University of Pittsburgh
 2 years med school
 3+ years grad school
 2 more years med school
 And then residency…
Roadmap



The topic: interfaces between the nervous system and
electronic devices
Why?
 What could they do for us?
 Do we really need that?
How?
 What problems do we have to solve?
 What techniques have been tried?
 What will we do next?
Nerve Chips: Why?

What could we do if we could tap into neural signals?




Route them around dead or damaged tissue
Replace missing sensory data
Control artificial limbs and organs (or anything
else that can be run by a computer…)
But even better yet….
P Heiduschka and S Thanos, 1998
Y Matsuoka, 2001
Nerve Chips: Why?

…expand human capabilities to the limits of human
imagination
Do We Really Need That?


Neurological disorders cost $250 billion/yr in USA
 Acute care, rehab, inability to work, long-term care
 Stroke, injuries, birth defects, diabetes,
Alzheimer’s, Parkinson’s, multiple sclerosis…
 No real cure for any of these
 Prosthetics exist, but hard to control
 No good sensory prosthetics (except hearing)
Would you like to…
 …see with better accuracy, even in the dark?
 …control your environment with a thought?
 …experience otherwise-impossible sensations?
Neuroanatomy in a Nutshell
What Do We Have to Do?







Get our interface into the body
Keep the body from attacking and rejecting the chip
Get close to the target nerve cells
Transmit electrical current to the targets
 Don’t transmit current to non-target cells
 Don’t harm the nerve with too much current
Record signals from the targets
 Try to separate out the voices of single cells
Do all this to thousands of cells at the same time
Adapt to the body changing over time
How Do We Do It? (1)
Nerve Cuff






Flexible cuff wrapped
around a whole nerve
Mechanically stable
Not very selective
Causes muscle fatigue
Can’t use in brain
Still a popular method
because it’s simple and
stable
P Heiduschka and S Thanos, 1998
How Do We Do It? (2)
Sieve Electrode



L Wallman et al., 1999

Axons of a cut nerve
regenerate through
holes in silicon chip
Lets us talk to
individual axons
We either have to wait
for a nerve to get cut or
cut it ourselves
Not in the clinic yet, but
soon…
How Do We Do It? (3)
Microelectrode Array






Array of tiny conducting
spikes
Can stick it anywhere in
the nervous system
Can’t be sure every
spike will hit a cell
PJ Rousche and RA Normann, 1998
Can damage tissue
Some clinical trials
ongoing
Versions of this let you
do some semi-cool
things with animals
What Can We Do Now? (1)

Cochlear Implants (hearing prosthesis)
 Pick up speech sounds with a microphone
 Filter digitally to reduce noise
 Pass to electrode array in cochlea (inner ear)
What Can We Do Now? (2)

Functional Electrical Stimulation (FES)
 Electrical stimulators similar to nerve cuff
 Implant near or inside key muscles
 Stimulation controlled by patient
commands (remote control device)
 Coordinated stimulation programs to
produce hand grasp, walking, etc.
 Can also trigger stimulation from sensors
What Can We Do Now? (2)
What Can We Do Now? (2)

Videos of FES Application: Correcting Foot Drop
What Can We Do Now? (3)






Visual prosthesis
Camera on glasses
Video sent to belt-pack
computer for
processing
10x10 electrode array
on the surface of visual
cortex
Actual result: 3-5
specks of light
(“phosphenes”)
Can read big text,
navigate in some
environments
What Can We Do Now? (4)






Multielectrode arrays to control
animal behavior
RoboRat (SUNY)
 Electrodes in “whisker” part of
brain indicate direction
 Electrodes in “pleasure” center
reward for correct behavior
RoboRoach (Tokyo University)
 Antennae replaced by electrode
Note large electronic backpack
required for each case
Effect wears off as animal adapts to
the stimuli
Any social/ethical implications?
What’s Still Missing?




All of these still use pretty big currents
 Hurts the cells, rapidly fatigues the muscles if
stimulating them directly
Need to be talking to a lot more cells to get true
biological precision and resolution
Only one (sieve electrode) is really specific for
individual cells
Can always use more mechanical stability and
biocompatibility
The Next Step?



Make a chip that
has living neurons
built into it
Use those living
cells as your
connection to the
patient
Nothing is better at
talking to neurons
than other
neurons…
How Do We Get There?






Nanotechnology – design of new “impossible” materials
Electrode coatings that contain brain molecules, “trick” cells
into acting like electrode is part of brain
Polymer chains that can enter the cell
Conductive polymer chains that place your electrode inside a
cell without hurting it
Components that “self-assemble” through chemical forces
Other crazy stuff I haven’t thought of yet
Thanks



Advisors
 Kaigham Gabriel (ECE, Robotics)
 Yoky Matsuoka (MechE, Robotics)
 Victor Weedn (MBIC)
Sources of Money
 NIH training grant T32N507433-03
 Department of the Army (NDSEG Fellowship)
 Paralyzed Veterans of America
Inspiration
 Dr. Joe Rosen
 You