Information Technology Reference
In-Depth Information
Results from our study agreed with clinical observations that patient 1 had the
least severe ULD and patient 3 had the most severe one. The highest coupling
strength in brain cortical networks was found in patient 3 before and after treat-
ment. However, patient 3 received 50 points on the UMRS before treatment and 66
after treatment due to the fact that patient 3 was able to engage in more testing after
treatment.
Although the results indicate that the mutual information and nonlinear interde-
pendencies measures could be useful in determining the treatment effects for pa-
tients with ULD, their limitations must be mentioned. It has been reported that it
is necessary to take into account the interdependence between thalamus and cor-
tex [33, 32]. By applying Granger causality in animal studies, Sitnikova et al. sug-
gested that onset of spike and wave discharges was associated with a rapid and sig-
nificant increase of coupling strength between frontal cortex and thalamus in both
directions. Furthermore, the strength of the thalamus to cortex coupling remained
constantly high during seizures [31]. The decoupling between frontal and occipital
cortical regions of our data after AED treatment may also be caused by decrease of
a driving force deep inside the brain. The effect of the treatment may thus reduce
the coupling strength between thalamus and cortex in ULD subjects.
It has been pointed out by several authors that nonlinear interdependence mea-
sures need to be applied with care [23]. For example, the embedding parameters
often play important roles for nonlinear analysis involving with state space recon-
struction. In this study, we used a false nearest neighbor algorithm and the mutual
information function for finding the embedding dimension
m
and delay
. However,
it is also known that there is no guarantee that these embedding parameters are the
optimal choices. Besides the intrinsic nonlinear properties, there are other sources
of noise underlying the real-world EEG recordings. Our strategy, in dealing with the
above potential drawbacks, is to fix the embedding parameters for the same set of
EEG recordings. By fixing the embedding parameters the underlying dynamics for
10.24 s SWD-free EEG recording, and therefore attractor, can consistently quantify.
At this point we would like to point out as a future research direction the need
to reproduce the same study using sleep EEG recordings and confirm if the number
and the distribution of decoupled electrode sites are the same and independent of
the state of vigilance. Also to prove the usefulness of the proposed study, a larger
patient population is needed.
Ď„
Acknowledgments
This work was partially supported by North Florida Foundation for Research
and Education, Inc. North Florida/South Georgia Veterans Health System 1601 SW Archer Rd.
(151), Gainesville, FL 32608.
References
1. Aarabi, A., Wallois, F., Grebe, R. Does spatiotemporal synchronization of EEG change prior
to absence seizures? Brain Res
1188
,207-221 (2008)
2. Arnhold, J., Grassberger, P., Lehnertz, K., Elger, C.E. A robust method for detecting interde-
pendences: Application to intracranially recorded EEG. Physica D
134
, 419-430 (1999)
3. Cao, L. Practical method for determining the minimum embedding dimension of a scalar time
series. Physica D
110
(1-2), 43-50 (1997)
