of 13
www.sciencemag.org/content/
348/
6237/
906
/suppl/DC1
Supplementary
Material
s for
Decoding motor imagery from the posterior parietal cortex of a tetraplegic
human
Tyson Aflalo, Spencer Kellis, Christian Klaes, Brian Lee, Ying Shi, Kelsie Pejsa, Kathleen
Shanfield, Stephanie Hayes
-Jackson, Mindy Aisen, Christi Heck, Charles Liu, Richard A.
Andersen
*
*Corresponding author. E
-mail: andersen@vis.caltech.edu
Published
22 May
2015,
Science
348
,
906
(2015)
DOI:
10.1126/science.
aaa5417
This PDF file includes:
Materials and Methods
Figs. S1 to S
3
Captions for
Movies S1 to S
3
References
Other Supporting Online Material for this manuscript includes the following
:
(available at
www.sciencemag.org/content/348/6237/906/suppl/DC1)
Movies S1 to S
3
Supplementary Materials:
Materials and Methods
Figures S1
-S3
Movies S1-
S3
References (
23
-26)
Materials and Methods:
All procedures were approved by the California Institute of Technology, University of
Southern California, and Rancho Los Amigos Internal Review Boards.
Informed consent was
obtained from EGS after the nature of the study and possible risks were explained. Study
sessions occurred at Rancho Los Amigos National Rehabilitat
ion Center.
Behavioral Setup
. All tasks were performed with EGS
seated in his motorized wheel
chair. Tasks involved the use of an anthropomorphic robotic limb, a 47 in. LCD monitor, or a
combination of both. The robotic limb was bolted to a steel frame positioned in front of and
offset to the right of the subject’s chair with the shoulder mount at approximately eye level. The
arm was always positioned 48 inches from the subject’s body to maintain a safety zone. When
the robotic limb was not needed for the task, it was moved outside the subject’s field of view.
The LCD monitor was positioned approximately 184 cms from the subject’s eyes. Stimulus
presentation was controlled using the Psychophysics Toolbox (
23) for MATLAB
.
We used the Modular Prosthetic Limb (MPL), a robotic arm designed through the Johns
Hopkins University (JHU) Applied Physics Laboratory (APL). It was designed to approximate
the function of a human arm, and its size and ranges of motion were specified to m
atch those of a
human arm as closely as possible. The MPL has 17 degrees of freedom, and, fully extended, the
MPL is approximately 79 cm long. During prosthetic control, the full set of degrees of freedom
were constrained to either 2d or 3d control of the position of the hand. In some tasks, EGS
was
required to use the MPL to point to targets displayed on the LCD display. A manual calibration
session was used to register the coordinate frames of the display and MPL for these sessions.
Neural Recordings
. EGS
was implanted with two 96
-channel Neuroport arrays in
putative homologues of area AIP and Brodmann’s Area 5d (
Fig. S
1B
). Neural activity was
amplified, digitized, and recorded at 30KHz with the Neuroport neural signal processor (NSP)
(Fig. 2). The N
europort System, comprising the arrays and NSP, has received FDA clearance for
<30 days acute recordings; for purposes of this study we received FDA IDE clearance (IDE
#G120096) for extending the duration of the implant.
For online prosthetic control, w
e used the time of unsorted action potential threshold
crossings. In the Central software suite, thresholds for action potential detection were set at -
4.5
times the root-
mean
-square of the high-
pass filtered (250Hz cut
-off) full
-bandwidth signal. The
time of threshold crossings were transmitted in real-
time to MATLAB, counted in non-
overlapping 50ms time bins, and utilized by the decoding algorithm to drive the prosthetic
device. For online prosthetic control, a minimum firing rate of 2 Hz was enforced.
For offline
analysis, single and multiunit activity was sorted using Gaussian mixture modeling of the 2d
principal component projection of waveforms detected via threshold crossing.
2