Information Technology Reference
In-Depth Information
In the scientific literature, improvements to DBS have been suggested by a num-
ber of authors [106, 146, 134, 39]. In particular, Montgomery and Baker [106]
suggested that a future direction of DBS would be to incorporate the ability of
acquiring and decoding neurophysiological information “to compute the desired
action.” Also, using results from a mathematical model of interconnected phase os-
cillators, Tass [146] proposes a method of demand-controlled double-pulse stimu-
lation that would hypothetically enhance the effectiveness of DBS while reducing
the power consumption of a stimulator in the long term. In addition, Sanghavi [134]
and Feng et al. [39] propose methods for adaptively modifying stimulus param-
eters while seeking to minimize measures of brain activity in the vicinity of the
implant.
13.7.1 Demand-Controlled DBS
From a theoretical perspective, Tass established a stimulus methodology based on
a model of Parkinsonian brain activity [146, 147]. In particular, Tass simulated the
synchronized oscillatory behavior of the basal ganglia using a network of phase
oscillators. This method is as follows: given
N
oscillators with global coupling
strength
K
0 where the phase, stimulus intensity, and noise associated with the
j
th oscillator are
>
, respectively, the behavior of the
j
th oscillator and
its relation to other oscillators as well as the stimulus is shown in Equations (13.4),
(13.5), and (13.6). In particular, defining factors
S
j
(
Ψ
j
)
Ψ
j
,
I
j
, and
F
j
(
t
)
and
X
j
(
t
)
as
S
j
(
Ψ
j
)=
I
j
cos
(
Ψ
j
)
and
(13.4)
1: neuron
j
is stimulated
0: otherwise
(
)=
,
X
j
t
(13.5)
the rate of change of the
j
th phase oscillator is given by
N
k
=
1
sin
(
ψ
j
−
ψ
k
)+
X
j
(
t
)
S
j
(
ψ
j
)+
F
j
(
t
)
.
K
N
ψ
=
Ω
−
˙
(13.6)
Tass showed that the model in Equations (13.4), (13.5), and (13.6) is able to
generate patterns of both synchronized oscillatory firing and random nonoscillatory
behavior. Moreover, the network tends to remain in a synchronized oscillation until
a global stimulus is applied at time
t
0
so that
X
j
1 for all
j
.
Effective stimulation methods for suppression of abnormal burst activity in this
model, as reported by Tass, include low-amplitude high-frequency stimulation (20
times the burst frequency), low-frequency stimulation (equal to the burst frequency),
or a single high-amplitude pulse, with the high-amplitude pulse being the most ef-
fective when it is applied at the appropriate phase of each neuron. Furthermore, Tass
proposes a demand-controlled stimulation technique whereby the synchronicity
(
t
0
)=
