IEEE Signal Processing Magazine - January 2018 - 13

SPECIAL REPORTS
John Edwards

Signal Processing Powers Next-Generation Prosthetics
Researchers investigate techniques that enable artificial limbs
to behave more like their natural counterparts

P

rosthetic limbs have improved sig-
nificantly over the past several years,
and signal processing has played a
key role in allowing these devices to
operate more smoothly and precisely
on command. Now, researchers are tak-
ing the next step forward by using signal
processing approaches and methods to
develop prosthetics that not only function
reliably and efficiently but give wearers
more natural control over artificial arms,
hands, and legs.
Researchers at London's Imperial
College, for instance, have developed a
prosthetic arm sensor technology that
detects signals transmitted by nerves in
the spinal cord. To control the prosthet-
ic, the wearer simply has to think about
controlling a phantom arm and imagine a
simple maneuver, such as pinching two
fingers together. The sensor technology
then interprets electrical signals sent from
the spine and uses them as commands.
Existing robotic prosthetic arms are
controlled by having the wearer twitch
remaining muscles in his or her shoul-
der or arm. The new approach detects
signals from spinal motor neurons in
parts of the body that were left undam-
aged by the amputation. A motor neu-
ron is a nerve cell is located in the spinal
cord with axons that project outside
the spinal cord to directly control body
muscles. The new method allows addi-
tional signals to be detected by sensors
Digital Object Identifier 10.1109/MSP.2017.2762821
Date of publication: 9 January 2018

1053-5888/18©2018IEEE

connected to the prosthetic, making the
recording electrical activity produced by
artificial limb more functional, explains
skeletal muscles, is central to the project.
project lead Dario Farina, neurorehabili-
"The core of the approach is the blind
tation engineering chair at Imperial Col-
source separation of high-density EMG
lege's Department of Bioengineering.
signals to identify the hidden neural
"We interface nerves by surgically
components," Farina explains. "Signal
directing them to target muscles above
processing is, therefore, the most impor-
the amputation and using the muscles
tant discipline in this research." In addition
as 'bioscreens' of the nerve activity,"
to blind deconvolution, signal processing
Farina says. When the nerves connect to
is also needed to map the decoded sources
muscles via reinnervation, each action
into commands to the prosthesis.
potential traveling
"We acquire high-
from the nerve to the
density
EMG signals
Now, researchers are
muscle elicits an action
from reinnervated
taking the next step
potential in the inner-
muscles," Farina says.
forward by using signal
vated muscle fibers.
The signals are a spa-
processing approaches
"We then decode the
tial sampling of the
and methods to develop
electrical muscle fiber
electrical potential
activity in order to id-
generated by the mus-
prosthetics that not
entify the command-
cle fibers. The elec-
only function reliably
ing nerve activity,"
trical activity of the
and efficiently but give
Farina says.
muscle fibers is deter-
wearers more natural
Farina notes that
mined by the neural
control over artificial
the surgical redirection
flow from the spinal
arms, hands, and legs.
of nerves to muscles
motor neurons. The
has been previously
high-density EMG
investigated by other research teams and
signals are filtered in time and space to
is referred to as targeted muscle reinner-
reduce power line interferences, arti-
vation. "We have added to this technique
facts, and other noise sources. "Then
the exact decoding of the individual nerve
the multichannel signals are spatially
fiber activity by deconvolution of the
whitened to remove their correlation,"
muscle electrical signals to identify the dis-
Farina says.
charge timings of individual spinal motor
Source identification of the EMG
neurons," Farina says. "This is done by
signals is approached in multiple ways.
blind source separation approaches ad-
"If the EMG recordings are suffici-
apted to the specific signal model."
ently selective, such as in the case of
The processing of electromyographic
implanted sensors, the spiking of indi-
(EMG) signals, used for evaluating and
vidual motor neurons can be extracted
IEEE SIGNAL PROCESSING MAGAZINE

|

January 2018

|

13
Table of Contents for the Digital Edition of IEEE Signal Processing Magazine - January 2018
