Biomedical Engineering Reference
In-Depth Information
TaBlE 5.1:
Subsets of neural channels
SUBSET NUMBER
EXPLANATION
1
All 104 neural channels
2-5
Different combinations of neural channels with statistically
significant correlations between arm movement/nonmovement [3]
6
Neural channels determined to be significant in artificial neural
network mappings of neural activity to 3D arm movement
7
51 neural probe channels
a
a
The 104-channel neural data through spike sorting of data from 51 neural probes.
In terms of the final trajectory reconstruction, each HMM will be gating a linear model for
the particular pieces of the trajectory where it wins the competition (i.e., rest or movement). Each
linear model is a multiinput-multioutput finite impulse response (FIR) filter that is trained using
normalized least mean square:
η
γ
e
( )
( )
winne
r
n X n
(5.57)
W
(
n
+ =
1
)
W
( )
n
+
2
winner
winne
r
+
X n
( )
TaBlE 5.2:
Results from varying the different parameters and neural subsets
L
N
n
SUBSET NO.
STATIONARY %
MOVING %
1
82.1
85.6
8
7
7
4
83.5
87.5
128
10
6
5
84.0
87.5
256
10
6
6
75.6
81.3
256
10
4
7
87.3
86.1
256
10
5
