Biomedical Engineering Reference
In-Depth Information
1, where [
x
] denotes the
integer part of
x
. To calculate the probability
p
n
(
m
) within each window
W
(
n
;
w
;
Here
Δ≤
w
is the sliding step and
n
=
0, 1, ..., [
n
/
Δ
]
−
w
+
Δ
),
M
we introduce intervals
W
(
n
;
w
;
Δ
)
=
I
m
.The probability
p
n
(
m
) of the sampled sig-
m
=
1
nal belonging to the interval
I
m
is the ratio of the number of the signals found within
interval
I
m
and the total number of signals in
W
(
n
;
w
;
Δ
). The value of
SE
(
n
)is
defined using
p
n
(
m
) as
M
=−
∑
()
()
()
SE n
p
m
log
2
p
m
(7.5)
n
n
m
1
7.3.1 Information Quantity
Although it is common to use the distribution of signal amplitudes to calculate the
entropy, there is no reason why other signal measures could not be employed. For
example, Fourier coefficients reflect the signal power distribution, whereas the
wavelet coefficients reflect the different signal scales, roughly corresponding to
coarse and fine time scales or correspondingly low- and high-frequency bands.
Instead of calculating entropy of the amplitude of the sampled signals, entropy of
the wavelet coefficients of the signal may be calculated to get an estimate of the
entropy in different wavelet subbands. Wavelet analysis decomposes the signal into
its different scales, from coarse to fine. Wavelet analysis of the signal is carried out
to decompose the EEG signals into wavelet subbands, which can be interpreted as
frequency subbands.
We calculate the IQ information theoretic analysis on the wavelet subbands.
First the discrete wavelet transform (DWT) coefficients within each window are
obtained as
WC
(
r
;
n
;
w
;
)]. The wavelet coefficients are
obtained from the DWT, and the IQ is obtained from the probability distribution of
the wavelet coefficients as follows:
Δ
)
=
DWT[
W
(
n
;
w
;
Δ
M
=−
∑
()
()
()
wc
wc
IQ n
p
m
log
2
p
m
(7.6)
n
n
m
1
where
p
n
(
m
) is an estimated probability that the wavelet-transformed signal belongs
to the
m
th bin where
M
is the number of bin.
We calculate IQ from a temporal sliding window block of the EEG signal as
explained earlier. Figure 7.4 shows the IQ trend plots for two experimental subjects.
IQ trends accurately indicate the progression of recovery after CA injury. The time
trends indicate the changing values of IQ during the various phases of the experi-
ments following injury and during recovery. The value of these trends lies in com-
paring the differences in the response to hypothermia and normothermia. There are
evident differences in the IQ trends for hypothermia versus normothermia. Hypo-
thermia improves the IQ levels showing quicker recovery under hypothermia and
over the 72-hour duration. The final IQ level is closer to the baseline (hatched line)
under hypothermia. These results support the idea of using IQ trends to monitor
brain electrical activity following injury by CA.
Search WWH ::
Custom Search