February 23, 2011

16:19

0218-348X

S0218348X1100521X

Fractals, Vol. 19, No. 1 (2011) 113123

c World Scientic Publishing Company


DOI: 10.1142/S0218348X1100521X

DIFFERENT ANESTHESIA IN RAT

INDUCES DISTINCT INTER-STRUCTURE
BRAIN DYNAMIC DETECTED BY HIGUCHI
FRACTAL DIMENSION
SLADJANA SPASIC,, SRDJAN KESIC, ALEKSANDAR KALAUZI
and JASNA SAPONJIC,
Department for Life Sciences, Institute for Multidisciplinary Research
University of Belgrade, Kneza Viseslava 1, 11 030 Belgrade, Serbia
Department of Neurobiology
Institute for Biological ResearchSinisa Stankovic
University of Belgrade, Despot Stefan Blvd. 142
11060 Belgrade, Serbia
sladjana@imsi.rs; sladjana@ibiss.bg.ac.rs
jasnasap@ibiss.bg.ac.rs; jasnasaponjic@yahoo.com
Received August 25, 2009
Revised May 12, 2010
Accepted November 16, 2010

Abstract
The complexity, entropy and other non-linear measures of the electroencephalogram (EEG),
such as Higuchi fractal dimension (FD), have been recently proposed as the measures of anesthesia depth and sedation. We hypothesized that during unconciousness in rats induced by the
general anesthetics with opposite mechanism of action, behaviorally and poligraphically controlled as appropriately achieved stable anesthesia, we can detect distinct inter-structure brain
dynamic using mean FDs. We used the surrogate data test for nonlinearity in order to establish
the existence of nonlinear dynamics, and to justify the use of FD as a nonlinear measure in the
,

Corresponding authors.

113

February
23,
S0218348X1100521X

114

2011

16:19

0218-348X

S. Spasic et al.

time series analysis. The surrogate data of predened probability distribution and autocorrelation properties have been generated using the algorithm of statically transformed autoregressive
process (STAP). FD then is applied to quantify EEG signal complexity at the cortical, hippocampal and pontine level during stable general anesthesia (ketamine/xylazine or nembutal
anesthesia).
Our study showed for the rst time that global neuronal inhibition caused by dierent
mechanisms of anesthetic action induced distinct brain inter-structure complexity gradient in
Sprague Dawley rats. EEG signal complexities were higher at cortical and hippocampal level in
ketamine/xylazine vs. nembutal anesthesia, with the dominance of hippocampal complexity. In
nembutal anesthesia the complexity dominance moved to pontine level, and ponto-hippocampocortical decreasing complexity gradient was established. This study has proved the Higuchi
fractal dimension as a valuable tool for measuring the anesthesia induced inter-structure EEG
complexity.
Keywords: Higuchi Fractal Dimension;
Ketamine/Xylazine; Nembutal; Anesthesia.

1. INTRODUCTION
During last decade, several nonlinear measures
such as nonlinear correlation index, Higuchi fractal dimension (FD), Shannon entropy, approximate
entropy and Lempel-Ziv complexity, have been successfully introduced in EEG analysis to investigate neuronal dynamic behavior.14 Investigation
of the brain dynamics by nonlinear analysis of the
EEG signals is used to dene the depth of sedation under anesthesia,5,6 to compare the proprieties
of the specic spontaneous or anesthesia induced
EEG waveforms,7 and to discriminate ictal (seizure)
from pre-ictal (before seizure) brain activity or critical brain site for seizure.8 The nonlinear measures
also dened the insomnia subtypes on the bases of
characterized daytime cortical activation in normal
sleepers versus chronic insomniacs.9 Since actual
living biological systems are not stable, and the
system complexity varies with time, it is possible
to distinguish dierent states of the system by the
fractal dimension, or to dene dierent complexity
between the structures within the same biological
system such as the brain.8,10,11
In this study, by using Higuchi fractal dimension, as a measure of the EEG signal complexity,
we compared the brain complexity at the distinct
brain levels (sensorimotor cortex, hippocampus and
pons), during stable anesthesia induced by different anesthetics (ketamine/xylazine or nembutal
anesthesia), and appropriately achieved for operative and experimental procedure in rat. Stability of anesthesia were estimated on the bases of
the observed loss of consciousness, muscle atonia,

Nonlinearity

in

EEG;

Surrogate

Data;

absence of the tail-pinch, ear-pinch (analgesia) and

corneal reexes, and on the bases of poligraphic
recording during experimental procedure, including
regular breathing pattern.12,13
Two main objectives underlie this study. First,
we compared the EEG signal complexities under
ketamine/xylazine or nembutal anesthesia at dierent brain levels (cortical, hippocampal and pontine
EEGs) using Higuchi fractal dimension. Second, the
EEG complexities within the distinct brain structures were compared in ketamine versus nembutal
anesthesia.
We hypothesized that it is possible to detect different complexity of the EEG signals at dierent
levels within the rat brain during stable anesthesia, caused by two anesthetics with opposite mechanisms of action (ketamine/xylazine or nembutal
anesthesia).

2. MATERIALS AND METHODS

2.1. EEG Data: Experimental
Animals, Operative Procedure,
Data Recording and
Acquisition
Experiments were performed in 17 male Sprague
Dawley rats, weighing 200300 g prior to surgery,
maintained on a 12 h light-dark cycle, and housed
at 25 C with free access to food and water. Principles for the care and use of laboratory animals in
research were strictly followed, as outlined by the
Guide for the Care and use of Laboratory Animals

February
23,
S0218348X1100521X

2011

16:19

0218-348X

Brain Structure Complexity Patterns in Anesthesia

(National Academy of Sciences Press, Washington,

DC, 1996).
Rats were anesthetized with a combination of
80 mg/kg ketamine (Abbott Laboratories, North
Chicago, IL), and 5 mg/kg xylazine (Phoenix Scientic Inc. St Joseph, MO) (8/17) or with 50 mg/kg
nembutal (9/17) given by intraperitoneal injection.
After a surgical plane of anesthesia was achieved
(controlled by absence of the tail- or ear-pinch and
corneal reexes), rats were placed in the stereotaxic
apparatus (David Kopf Inst., model 962 A Tujunga,
CA). Following the electrodes implantation, bilateral referential EEGs were recorded using stainless
screws in the parietal cortex (P: 2.5; R/L: 2; V:
1 to bregma), and teon coated wires (Medwire,
MT. Vernon, NY) in the CA1 region of hippocampus (P: 3.8; R/L: 1.7; V: 2.7 to bregma). A bipolar twisted wire electrode with uninsulated tips of
1 mm was stereotaxically targeted into right pedunculopontine tegmental nucleus (PPT) to record the
pontine EEG (P: 8; R: 1.8; V: 7 to bregma).
During each experimental protocol, we performed
an 8-channel recording: (1) respiratory movements
monitored by a piezoelectric strain gauge (InfantPed Sleepmate Technologies, Mislothian, VA); (2)
an injection marker signal obtained from MPPI-2;
(3) EEG from left sensorimotor cortex; (4) EEG
from right sensorimotor cortex; (5) left hippocampal EEG; (6) right hippocampal EEG; (7) right pontine EEG; and (8) genioglossal EMG. The details of
this method are more fully explained elsewhere.12,13
After conventional amplication and ltering (0.1
50 Hz band-pass; Grass Model 12, West Warwick)
the analog data were digitized (sampling frequency
100/s) and recorded using Brain Wave for Windows
software (Datawave Systems, Longmont, CO).
In this study we used 180 s of all EEG records
during stable control anesthetized condition: (1)
EEG from left sensorimotor cortex; (2) EEG from
right sensorimotor cortex; (3) left hippocampal
EEG; (4) right hippocampal EEG; and (5) right
pontine EEG.

2.2. STAP Algorithm

The application of nonlinear methods in time series
analysis, like fractal or other analysis has to be
justied by establishing nonlinearity in the time
series.14 We have used the algorithm of statically
transformed autoregressive process (STAP) to generate the surrogate data with detailed description

115

in Kugiumtzis, 2002.15 For the given time series, we

have generated the surrogate data sets, and calculated the Higuchis fractal dimension for the original
and surrogate data. Finally, we have tested where
the FD of the data lies in the context of the surrogate results. When the original data yields the
FD statistical signicant dierence from FDs of the
surrogate data sets, we can have more condence
that original EEG signals arises from a nonlinear
process. STAP algorithm generates the surrogate
data as realizations of a suitable statically transformed autoregressive process. It uses a typical realizations approach and attempts to build a proper
autoregressive (AR) model in order to generate data
that match two conditions. The rst, surrogate time
series preserves exactly the marginal probability
density function of the original time series and the
second the STAP algorithm nds analytically
the autocorrelation of the surrogate and original
time series for sucient range of lags . The STAP
algorithm is applied using MATLAB software by
Kugiumtzis. We used STAP software with following free parameters: m = 5 as the degree of the
polynomial to approximate the sample transform,
p = 50 as an order of the AR(p) model to generate
the Gaussian time series, M = 50 is a number of
surrogates to be generated, max = p = 50 is maximum lag to be used in the comparison of autocorrelations, and K = M = 50 is a number of repetitions
to be made to nd the best AR model.

2.3. Higuchi Fractal Dimension

Fractal analysis was performed by estimating fractal dimension values of the cerebral cortical, hippocampal and pontine EEG signals using Higuchis
algorithm.16 Fractal dimension is a nonlinear measure of signal complexity in time domain. It is
very simple and useful method for assessment
nonlinearity of signal without reconstruction a
strange attractor.17 Briey, EEG was analyzed
in time sequences x(1), x(2), . . . , x(N ) and it was
constructed in a new self-similar time series
Xkm as:
Xkm : x(m),

x(m + k),

x(m + 2k), . . . , x(m + int[(N k)/k]k)

for m = 1, 2, . . . , k where m is initial time; k =
2, . . . , kmax , where k is time interval, kmax is a free
parameter, and int(r) is integer part of the real

February
23,
S0218348X1100521X

116

2011

16:19

0218-348X

S. Spasic et al.

number r. The length Lm (k) was computed for each

of the k time series or curves Xkm .

int[ Nm ]
1 k
|x(m + ik)
Lm (k) =
k
i=1

x(m + (i 1)k)|

N 1

 ,
int N m
k
k

where N is the length of the original time series X

and (N 1)/{int[(N m)/k]k} is a normalization
factor. Lm (k) was averaged for all m forming the
mean value of the curve length L(k) for each k =
2, . . . , kmax as
k
Lm (k)
.
L(k) = m=1
k
An array of mean values L(k) was obtained and the
FD was estimated as the slope of least squares linear
best t from the plot of ln(L(k)) versus ln(1/k):
F D = ln(L(k))/ ln(1/k).
After preliminary tests, our results conrm that the
FD is independent of the length of the window, at
least for N > 100, but much more dependent of
the parameter kmax . So, we chose the parameter
N = 500 equivalent to an epoch of 5 s duration and
the parameter kmax = 8. We used the parameter
kmax = 8 on the basis of our former study18 of the
optimum choice for kmax value. EEG signals were
divided into 36 epochs (windows). FD values were
calculated for each epoch, without overlap. Individual FD values were averaged across all epochs
for the signal. Higuchis algorithm is applied using
MATLAB software that has been validated in our
previous publications.10,11,18,19
In order to facilitate the interpretation of FD values of the EEG signals, the FD value of smooth
curve (for example linear or a low frequency sine
wave) was estimated to be 1. The FD of random
white noise was estimated to be 2.

2.4. Fourier Amplitude Spectra

We also calculated Fourier amplitude spectra on
each 5 s epoch during overall 180 s of stable control anesthetized condition for each animal and for
all ve recorded structures as described before. For
each type of anesthesia and each recorded structure we averaged normalized amplitude spectra (for
nembutal anesthesia N = 9, for ketamine/xylazine
anesthesia N = 8).

2.5. Statistical Test for Nonlinearity

A general null hypothesis is that the signal is a
static nonlinear transformation s(xn ) of a linear
Gaussian process xn . This hypothesis is tested by
generating surrogate data. We have generated an
ensemble of 50 surrogate data sets for every cerebral, hippocampal and pontine EEG signal in order
to test the null hypothesis H0 that the original
rat biosignals are generated by linear stochastic
process undergoing static nonlinear transform s.
The surrogate data must have the same autocorrelation, and the same amplitude distribution as
the original EEG signals to represent H0 . The null
hypothesis H0 is rejected at the 0.05 signicance
level, under the assumption of Gaussian distribution of statistic q, if the discriminating statistic
q0 , from an applied nonlinear method on the original data set, is statistically dierent from statistics
q1 , . . . , qM of the M surrogates, for range of the free
parameters ( and m) for each statistic. As a discriminating statistics, we used the Higuchi fractal
dimension.
To compare the original and the surrogate series,
discriminative statistics, FD are calculated for both
series. We used one sample two-tail t-test. The
statistics of signicance S is
S(i) =

|FD0 (i) FDs (i)|

,
FD (i)

where FD0 (i) stands for the statistic on the original time series, FDs (i) stands for the average of the
statistic on the surrogate time series, and FD (i)
is the standard deviation of the statistic on the
surrogate time series, for each original signal i =
1, 2, . . . , 85. For = 0.05, the critical value of S
is 1.96. Therefore, the null hypothesis was rejected
when S > 1.96, at the 0.05 signicance level and
the original signal was considered as nonlinear.

2.6. Statistical Analysis of Data

We tested the dierences between the Higuchi fractal dimensions of EEG signals from the certain
brain structures of ketamine/xylazine or nembutal
anesthetized rats using one-factor Fridman ANOVA
with post-hoc Wilcoxon Matched Pairs Test. The
dierences between fractal dimension values of the
brain structures between groups (ketamin/xylazine
vs. nembutal anesthetized rats) we tested by onefactor Kruskal-Wallis ANOVA and post-hoc MannWhitney U Test.

February
23,
S0218348X1100521X

2011

16:19

0218-348X

Brain Structure Complexity Patterns in Anesthesia

3. RESULTS
We have calculated Higuchi fractal dimension of
the sensorimotor cortical, hippocampal and pontine EEG signals in dierent experimental conditions: in ketamine/xylazine or nembutal anesthesia.
Higuchi fractal dimension values were calculated for
all (40 + 45) signals in (8 + 9) rats, during stable
control anesthetized condition.

3.1. Surrogate Data Analysis

The evidences of nonlinearity (Fig. 1), with FD as
discriminative statistics, were found in 7/8 rats in
left cerebral cortical EEG, 8/8 rats in right cerebral
cortical EEG, 5/8 rats in left hippocampal EEG,
7/8 rats in right hippocampal EEG, and 6/8 rats
in pontine EEG in ketamine/xylazine anesthesia.
The evidences of nonlinearity were found in 8/9
rats in left cerebral cortical EEG, 8/9 rats in right
cerebral cortical EEG, 5/9 rats in left hippocampus, 8/9 rats in right hippocampus, and in 8/9 rats
in pontine EEG in nembutal anesthesia. Our null
hypothesis that the EEG series are generated by
the statically transformed autoregressive process,
AR(50), has been rejected in 73.68% cases of cerebral, hippocampal, and pontine EEG. Of course, the
fact that the null hypothesis can not be rejected on
the basis of the given evidence does not prove that
the dynamic originated from statically transformed
autoregressive process.

117

bases of the mean FD values (Fig. 2(c), Table 1).

Friedman ANOVA showed signicant structure
related complexity dierence (ChiSqr = 12.50,
N = 8, df = 4, p < 0.014).
The highest mean FD value is in the left hippocampus (Fig. 2(c)). The hippocampal complexity
is higher than cortical complexity on the same side
of the brain (Figs. 2(a)2(c); z 2.24; p 0.05).
Complexity of the left hippocampus is higher than
pontine complexity (z = 2.38; p = 0.02), and mean

(a)

(b)

3.2. Structure Related Complexity

Under Ketamine/Xylazine
Anesthesia
We evaluated intrastructural complexity of the rat
brain under ketamine/xylazine anesthesia on the

(c)

Fig. 1 Testing hypothesis of nonlinearity by Higuchi fractal

dimension as statistics. Results are depicted in percentages.

Fig. 2 Dierent brain inter-structure complexity gradient under ketamine/xylazine anesthesia ((b) left brain;
(a) right brain) in Sprague-Dawley rats, and the mean
value and standard deviation of the Higuchi fractal dimensions (mean FD and sd FD) of each brain structure in
ketamine/xylazine anesthesia (c) Cx cortex; Hipp hippocampus, PPT pedunculopontine nucleus within the pons.
Solid arrow statistically signicant order of complexity;
dashed arrow tendency of complexity order.

February
23,
S0218348X1100521X

118

2011

16:19

0218-348X

S. Spasic et al.

Table 1 Results of Wilcoxon Matched Pairs

Tests of the Interstructure FDs Dierences Under
Ketamine/Xylazine Anesthesia. Bold p-Values are
Signicant.
p-Level

Cx Right

Hipp Left

Hipp Right

Pons

Cx left
Cx right
Hip left
Hip right

0.40

0.03
0.03

0.12
0.05
0.07

0.48
0.33
0.02
0.07

FD of the left hippocampus shows a tendency to

be higher than the right hippocampal FD (z =
1.82; p = 0.07), while FD of the right hippocampus shows a tendency to be higher than pontine

FD (Fig. 2(b); Table 1; z = 1.82; p = 0.07). There

is no dierence between FDs of the left and right
cortex (z = 0.84; p = 0.40). Pontine complexity is
not dierent than complexity of both cortical sides
(z 0.98; p 0.48).
Table 1 indicates the p-values calculated pairwise
for the mean FDs of the EEG signals: cortex left,
cortex right, hippocampus left, hippocampus right,
and pons right.
Figures 3(a) and 3(b) depicts the corresponding averaged (N = 8) normalized EEG spectra
of 180 s of stable control condition per each structure within the same side of the brain (right or
left brain side) during ketamine/xylazine anesthesia

(a)

(b)

(c)

(d)

Fig. 3 Averaged normalized amplitude spectra of 180 s of EEG under ketamine ((a) left side, (b) right side; N = 8), and
nembutal ((c) left side, (d) right side; N = 9) anesthesia including pontine (dotted line), hippocampal (dashed line), and
cortical (black-solid line) EEG spectra that we analyzed using nonlinear measure of complexity Higuchi fractal dimension.

February
23,
S0218348X1100521X

2011

16:19

0218-348X

Brain Structure Complexity Patterns in Anesthesia

that we analyzed using nonlinear measure of complexity Higuchi fractal dimension. The characteristic 510 Hz frequency activity could be noticed
on both brain sides in each structure.

3.3. Structure Related Complexity

Under Nembutal Anesthesia
Friedman ANOVA showed signicant structure
related complexity dierence between the brain

(a)

(b)

119

structures of nembutal anesthetized

rats
(Chi Sqr = 19.60, N = 9, df = 4, p < 6 104 ).
Figure 4(c) depicts that the left hippocampal complexity was the highest. Like in ketamine/xylazine
anesthesia, the hippocampal complexity is higher
than cortical complexity of the same brain side
(z 2.67; p 0.40), and there is no dierence
between complexity of the left and right cortex
(z = 1.60; p = 0.11). Also, left hippocamal complexity is higher than pontine complexity (Figs. 4(a)
and 4(c); z = 2.19; p = 0.03). In contrast to
ketamine anesthesia, the pontine EEG signal complexity (Figs. 4(a) and 4(b)) was higher than the
cortical signal complexities (z = 2.55; p = 0.01),
and right hippocampal complexity (z = 1.95; p =
0.05). Although mean FD of the left hippocampus
is the highest like in ketamine anesthesia, in the
nembutal anesthetized rats the pontohippocampocortical complexity gradient was established within
the right brain the highest complexity in the pons
with decrement to cortical level (Fig. 4(b)).
Results of Wilcoxon Matched Pairs Test are
shown in (Table 2), and the p-value calculated pairwise for the EEG signals are depicted: cortex left,
cortex right, hippocampus left, hippocampus right,
and pons right.
The corresponding averaged (N = 9) normalized EEG spectra of 180 s of stable control condition per each structure within the same side of
the brain (right or left brain side) during nembutal
anesthesia that we analyzed using FD are depicted
on Figs. 3(c) and 3(d).

3.4. Structure Related Brain

Complexities Dierences in
Ketamine/Xylazine vs.
Nembutal Anesthesia
We evaluated complexity of the rat brain under
ketamine vs. nembutal anesthesia by Kruskal-Wallis
ANOVA of independent grouping variables. There
is a signicant dierence between complexity of the
(c)

Fig. 4 Dierent brain inter-structure complexity gradient

under nembutal anesthesia ((a) left brain; (b) right brain)
in Sprague-Dawley rats, and the mean value and standard
deviation of the Higuchi fractal dimensions (mean FD and
sd FD) of each brain structure (c) Cx cortex; Hipp hippocampus, PPT pedunculopontine nucleus within the pons.
Solid arrow statistically signicant order of complexity.
Characteristic ponto-hippocampo-cortical complexity gradient established within the right brain side is depicted (b).

Table 2 Results of Wilcoxon Matched Pairs Test of

the Interstructure FDs Dierences Under Nembutal
Anesthesia. Bold p-Values are Signicant.
p-Level

Cx Right

Hipp Left

Hipp Right

Pons

Cx left
Cx right
Hip left
Hip right

0.11

0.01
0.01

0.44
0.04
0.02

0.01
0.01
0.03
0.05

February
23,
S0218348X1100521X

120

2011

16:19

0218-348X

S. Spasic et al.
Table 3 Mean FD Values and Corresponding
Standard Deviations (sd FD) of Each Brain Structure in Ketamine/Xylazine and Nembutal Anesthesia.
Mean FD sd FD

Brain Structure

Ketamine/Xylazine
(a)

(b)

Cx left
Cx right
Hipp left
Hipp right
Pons

1.39
1.38
1.52
1.45
1.40

0.05
0.05
0.10
0.04
0.06

Nembutal
1.32
1.31
1.43
1.35
1.38

0.03
0.05
0.09
0.08
0.06

(z = 1.54; p = 0.14), as well as in pons (z =

1.15; p = 0.28) in dierent anesthesias.
All mean FD values and corresponding standard
deviations (sd FD) calculated for EEG signals of
dierent brain structures and for both anesthetized
conditions are shown in (Table 3).

4. DISCUSSION

(c)

Fig. 5 Comparison of the inter-structure brain complexity in ketamine vs. nembutal anesthesia ((a) left brain;
(b) right brain) in Sprague-Dawley rats, and the mean value
the Higuchi fractal dimensions (mean FD) of each brain structure (c) Cx cortex; Hipp hippocampus, PPT pedunculopontine nucleus within the pons. ** p 0.01.

brain structures in ketamine/xylazine versus nembutal anesthetized rats (H(9, N = 85) = 44.43; p <
105 ). Mann-Whitney U Test showed (Fig. 5(c))
that cortical complexity of the left (Fig. 5(a); z =
3.08; p < 9 104 ) as well as of the right side
(Fig. 5(b); z = 2.60; p < 7 103 ) was significantly higher in ketamine vs. nembutal anesthesia. Right hippocampal complexity was signicantly
higher (z = 2.60; p < 7103 ) in ketamine anesthesia. However, there is no statistically signicant differences between the left hipocampal complexities

We have shown for the rst time that at the same

level of anesthesia induced unconsciousness established by distinct anesthetics (ketamin/xylazine or
nembutal) in Sprague-Dawley rats, with appropriately achieved analgesia, atonia, areexia and
regular breathing pattern, there was still a significantly dierent brain inter-structural EEG complexity pattern. Generally, in ketamine/xylazine
anesthesia the brain complexity was higher at cortical and hippocampal level with respect to nembutal anesthesia (Figs. 5(a) left brain, 5(b) right
brain). During stable ketamine/xylazine anesthesia
there was a dominance of hippocampal complexity
(Fig. 2(a) left brain, 2(b) right brain), while in nembutal anesthesia the complexity dominance moved
to the pontine level, and ponto-hippocampo-cortical
complexity gradient was established (Fig. 4(a) left
brain, 2(b) right brain).
A major problem in anesthesia is to obtain
objective measures of anesthetic depth.20 MayerKress and Layne21 were the rst who questioned
whether nonlinear EEG analysis (using the correlation dimension) can be used to estimate anesthetic depth. Their results would seem to argue
against the use of this nonlinear quantier as the
simple measure of anesthetic depth. However, Watt
and Hamero22 concluded that correlation dimension may be of interest for characterization the
three states of consciousness namely, the correlation dimension decreases as the anesthetic depth

February
23,
S0218348X1100521X

2011

16:19

0218-348X

Brain Structure Complexity Patterns in Anesthesia

increases. More recently, correlation dimension,1,23

Shannon entropy24 and approximate entropy2 are
conrmed as useful. Some studies,4,6,7 have shown
that the Higuchis fractal dimension method may
be successfully applied for analysis of EEG signals
for monitoring the depth of anesthesia or sedation
in fentanyl and propofol anesthesia.
Our study evidenced that this is also an important issue for analysis and explanations of the
results that we get using any in vivo animal anesthetized model. Using Higuchi fractal dimension we
were able to detect distinct inter-structure complexity (distinct degree of neuronal activity), and to
uncover that the poligraphically and behaviorally
dened stable anesthetized state as equal during
experiment in both anesthesias, is not the same
with respect to degree of neuronal activity (degree
of inhibition) at dierent brain levels.
Ketamine/xylazine mixture and nembutal anesthesia are commonly used anesthetics for laboratory rats as well as in humans. Xylazine is an
2 adrenoreceptors agonist that is a very commonly used sedative and muscle relaxant. It is
well known that the ketamine and nembutal have
the opposite main mechanisms of action. While
ketamine antagonizes glutamatergic receptors (glutamate is one of the brain excitatory neurotransmitter) and suppresses excitation within the
brain, nembutal is the agonist of GABA-ergic
receptors (GABA is the main brain inhibitory
neurotransmitter) through which it potentiates
inhibition within the brain. Beside the main mechanisms of action, the opposite eects of ketamine
versus nembutal has been evidenced also on acetilcholine release in Sprague-Dawley rats. In contrast to ketamine which increases acetil-choline
release in hippocampus and frontal cortex, but
had no eect in the striatum, nembutal decreases
acetil-choline release in all structures.25,26 An
important role of ketamine in hippocampal acetilcholine release is explained through the evidence
of the highest density of NMDA receptors in rat
hippocampus.27 Ketamine is a unique anesthetic
that interacts with NMDA, opioid, monoaminergic and muscarinergic receptors and voltagesensitive Ca++ channels, but does not interact
with GABA receptors.28 The non-NMDA mechanism of ketamine action is through an increase
of noradrenaline and acetyl-choline, while nembutal does not have such eect.25,28,29,30 The characteristic post-anesthetic increase in noradrenaline
release caused by ketamine also might contribute

121

to psychomimetic eects of ketamine such as vivid

dreams, agitation and illusions.29 Generally, NMDA
receptor antagonism or GABAA receptor activation
has been considered as important target sites in the
general anesthesia induction.31 From the structure
related point of view the cerebral cortex has been
recognized as a target site of general anesthesia,32
while the medial prefrontal cortex as the most
important brain site for manifestation of emotional
behavior.33
Our results have proved in both anesthesias,
in terms of cortical EEG complexity, the cortex
as a main target of loss of consciousness in general anesthesia. Although in ketamine vs. nembutal anesthesia there was a general higher EEG
complexity at cortical and hippocampal level, in
ketamine anesthesia the dominant EEG complexity
still stays at the limbic cortical level (hippocampus), while in nembutal anesthesia the EEG complexity decrement order turns to start from the
pons through hippocampus to cortex. Although we
recorded pontine EEG only on the right brain side,
we assume that the same EEG complexity pattern gradient is established on the left brain side
under nembutal anesthesia. On the bases of normalized amplitude spectra shown on Fig. 3 the dierences in the inter-structure complexity in ketamine
vs. nembutal anesthesia were particularly recognized as more prominent presence of the amplitudes within 510 Hz frequency range in ketamine
induced anesthesia. Our results also indicate that
generalized potentiation of the GABA eects by
nembutal within the brain induced suppression
of this particular frequency range activity at all
brain levels. So far we can only speculate that
this generalized higher activity within this frequency range under ketamine anesthesia yields
generally higher brain complexity with respect to
nembutal anesthesia (Figs. 5(a) and 5(b)). Precise quantitative contributions of particular frequency bands to the detected level of FD could be
assessed if a dedicated series of simulations were
performed (possibly based on combining signals
obtained by dierent band-pass ltering of the original signal).
We have shown that global neuronal inhibition caused by dierent mechanisms of anesthetic
action (suppression of excitation through the block
of glutamate NMDA receptors by ketamine or
potentiation of inhibition through the activation
of GABA receptors by nembutal) induces distinct
inter-structure EEG complexity pattern. This study

February
23,
S0218348X1100521X

122

2011

16:19

0218-348X

S. Spasic et al.

has proved the Higuchi fractal dimension as a valuable tool for measuring the anesthesia induced
inter-structure EEG complexity.

ACKNOWLEDGMENTS
The authors thank Prof. Dimitris Kugiumtzis for
MATLAB software package for surrogate data generation. This work was supported by Serbian Ministry of Science and Technological Development
[Project No. 173022] and by NIH [grant AG16303].

REFERENCES
1. G. Widman, T. Schreiber, B. Rehberg, A. Hoeft and
C. E. Elge, Quantication of depth of anesthesia
by nonlinear time series analysis of brain electrical
activity, Phys. Rev. E 62 (2000) 48984903.
2. J. Bruhn, H. Ropcke and A. Hoeft, Approximate
entropy as an electroencephalographic measure of
anesthetic drug eect during desurane anesthesia,
Anesthesiology 92 (2000) 715726.
3. X. S. Zhang, R. J. Roy and E. W. Jensen, EEG
complexity as a measure of depth of anesthesia
for patients, IEEE Trans. Biomed. Eng. 48 (2001)
14241433.
4. W. Klonowski, E. Olejaczyk, R. Stepien, P.
Jalowiecki and R. Rudner, Monitoring the depth
of anaesthesia using fractal complexity method in
Complexus Mundi, Emergent Patterns in Nature,
ed. M. M. Novak (World Scientic, Singapore, 2006)
pp. 333342.
5. R. Ferenets, T. Lipping, S. Melto and S. Hovilehto,
Comparison of entropy and complexity measures for
the assesment of depth of sedation, IEEE Trans.
Biomed. Eng. 53 (2006) 10671077.
6. A. Anier, T. Lipping, S. Melt and S. Hovilehto,
Higuchi fractal dimension and spectral entropy as
measures of depth of sedation in intensive care
unit, in Proceedings of the 26th Annual International
Conference of the IEEE EMBS (San Francisco, Ca,
USA, 2004) pp. 526529.
7. R. Ferenets, T. Lipping and P. Suominen, Comparison of the properties of EEG spindles in sleep and
propofol anesthesia, in Proceedings of the 28th IEEE
Annual International Conference (New York City,
USA, 2006) pp. 63566359.
8. P. Paramanathan and R. Uthayakumar, Application of fractal theory in analysis of human electroencephalographic signals, Comput. Biol. Med. 38
(2008) 372378.
9. E. Kronholm, J. Virkkala, T. K
arki, P. Karjalainen,
H. Lang and H. H
amal
ainen, Spectral power and
fractal dimension: methodological comparison in a
sample of normal sleepers and chronic insomniacs,
Sleep Biol. Rhythms 5 (2007) 239250.

c, G. Grbic and L. J.
10. S. Spasic, A. Kalauzi, M. Culi
Martac, Fractal analysis of rat brain activity after
injury, Med. Biol. Eng. Comput. 43 (2005) 345348.
c, G. Grbic, L. J. Martac, S.
11. S. Spasi c, M. Culi
Sekulic and D. Mutavdzic, Spectral and fractal analysis of cerebellar activity after single and repeated
brain injury, Bull. Math. Biol. 70 (2008) 12351249.
12. J. Saponjic, M. Radulovacki and W. D. Carley,
Injection of glutamate into pedunculopontine
tegmental nuclei of anesthetized rat causes respiratory dysrhythmia and alters EEG and EMG power,
Sleep Breath 9 (2005) 8291.
13. J. Saponjic, M. Radulovacki and W. D. Carley, Modulation of respiratory pattern and upper airway muscle activity by the pedunculopontine tegmentum:
role of NMDA receptors, Sleep Breath 10 (2006)
195202.
14. T. Schreiber and A. Schmitz, Surrogate time series,
Physica D 142 (2000) 346382.
15. D. Kugiumtzis, Statically transformed autoregressive process and surrogate data test for nonlinearity,
Phys. Rev. E 66 (2002) 025201.
16. T. Higuchi, Approach to an irregular time series on
the basis of the fractal theory, Phys. D 31 (1988)
277283.
17. A.
Accardo,
M
Anito,
M.
Carrozzi
and F. Bouquet, Use of the fractal dimension for
the analysis of electroencephalographic time series,
Biol. Cybern. 77 (1997) 339350.
c, G. Grbic and L. J.
18. S. Spasic, A. Kalauzi, M. Culi
Martac, Estimation of parameter kmax in fractal
analysis of rat brain activity, Ann. NY Acad. Sci.
1048 (2005) 427429.
c, G. Grbic and
19. A. Kalauzi, S. Spasic, M. Culi
L. Martac, Consecutive dierences as a method of
signal fractal analysis, Fractals 13 (2005) 283292.
20. S. J. Stam, Nonlinear dynamical analysis of EEG
and MEG: Review of an emerging eld, Clin. Neurophysiol. 116 (2005) 22662301.
21. G. Mayer-Kress and S. P. Layne, Dimensionality of
the human electroencephalogram, Ann. N. Y Acad.
Sci. 504 (1987) 6286.
22. R. C. Watt and S. R. Hamero, Phase space electroencephalography (EEG): a new mode of intraoperative EEG analysis, Intern. J. Clin. Mon. Comp.
5 (1988) 313.
23. P. L. C. Van den Broek, C. M. Van Rijn, J.
Van Egmond, A. M. L. Coenen and L. H. D. J.
Booij, An eective correlation dimension and burst
suppression ratio of the EEG in rat. Correlation
with sevourane induced anaesthetic depth, Eur. J.
Anaesth. 23 (2000) 391402.
24. J. Bruhn J, L. E. Lehmann, H. R
opcke, T. W. Bouillon and A. Hoeft, Shannon entropy applied to the
measurement of the electroencephalographic eects
of desurane, Anesthesiology 95 (2001) 3035.

February
23,
S0218348X1100521X

2011

16:19

0218-348X

Brain Structure Complexity Patterns in Anesthesia

25. K. Sato, J. Wu, T. Kikuchi, Y. Wang, I. Watanabe and F. Okumura, Dierential eects of ketamine
and pentobarbitone on acetylcholine release from
the rat hippocampus and striatum, Br. J. Anaesth.
77 (1996) 381384.
26. T. Kikuchi, Y. Wang, H. Shinbori, K. Sato and
F. Okumura, Eects of ketamine and pentobarbitone on acetylcholine release from the rat frontal
cortex in vivo, Br. J. Anaesth. 79 (1997) 128130.
27. D. M. Brockmeyer and J. J. Kendig, Selective eects
of ketamine on amino acid-mediated pathways in
neonatal rat spinal cord, Br. J. Anaesth. 74 (1995)
7994.
28. T. Kubota, N. Anzawa, K. Hirota, H. Yoshida,
T. Kushikata and A. Matsuki, Eects of ketamine
and pentobarbital on noradrenaline release from the
medial prefrontal cortex in rats, Can. J. Anesth. 46
(1999) 388392.
29. T. Kubota, K. Hirota and H. Yoshida, Eects of
sedatives on noradrenaline release from the medial

30.

31.

32.

33.

123

prefrontal cortex in rats, Psychopharmacology 146

(1999) 335338.
T. Kubota, K. Hirota, N. Anzawa, H. Yoshida,
T. Kushikata and A. Matsuki, Physostigmine antagonizes ketamine-induced noradrenaline release from
the medial prefrontal cortex in rats, Brain Res. 840
(1999) 175178.
J. G. Bovill, Mechanisms of intravenous anesthesia, in Textbook of Intravenous Anesthesia, ed.
P. F. White (Williams & Wilkins, Baltimore, 1997)
pp. 2746.
A. Angel and D. A. Gratton, The eect of anaesthetic agents on cerebral cortical responses in the
rat. Br. J. Pharmacol. 76 (1982) 541549.
J. Korf, G. K. Aghajanian and R. H. Roth, Increased
turnover of norepinephrine in the rat cerebral cortex
during stress: role of locus coeruleus, Neuropharmacology 12 (1973) 933938.