Clinical Neurophysiology 112 (2001) 38±45

www.elsevier.com/locate/clinph

Investigation of non-linear properties of multichannel EEG in the early

stages of Parkinson's disease
Laurent Pezard a, Robert Jech b, EvzÏen RuÊzÏicÏka b,*
a
Laboratoire de Neurosciences Comportementales, Universite Rene Descartes, 45 rue des Saints-PeÁres, F-75270 Paris Cedex 06, France
b
Clinic of Neurology, Charles University, KaterÏinska 30, CZ-120 00 Praha 2, Czech Republic
Accepted 25 October 2000

Abstract
Objectives: Modi®cations of brain activity in the early stages of Parkinson's disease (PD) are dif®cult to detect using electroencephalo-
graphy (EEG) signals and are often biased by l-DOPA treatment. We compare here the performances of both linear and non-linear methods
in differentiating EEG of l-DOPA naive PD patients from that of control subjects.
Methods: Resting multichannel EEG (20 electrodes, 30 s epochs) of 9 patients with PD in Hoehn and Yahr stages 1±2 (4 women, 5 men,
mean age 54.3 years, range 48±63 years) were compared with those of 9 control subjects (7 women, two men, mean age 51.3 years, range 43±
61 years). The following measurements were computed: u-, a- and b-band relative powers constituted the linear indices; localized entropy,
slope asymmetry and number of non-linear EEG segments constituted the non-linear indices.
Results: In the case of linear quanti®cation, only a decrease in the b-band was observed for patients. Signi®cant non-linear structures were
observed in our EEG data. Non-linear quanti®ers demonstrate an increase in entropy and in the number of non-linear EEG segments for the
patients.
Conclusions: Changes in EEG dynamics observed here in l-DOPA naive PD patients may represent early signs of cortical dysfunction
produced by subcortical dopamine depletion. q 2001 Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Parkinson's disease; EEG; Spectral analysis; Non-linear dynamics; Entropy; Surrogate data

1. Introduction motor and non-motor dysfunction (Agid et al., 1987; Pillon

et al., 1989). l-DOPA treatment may be another source of
Parkinson's disease (PD) is characterized by a motor bias as demonstrated by studies of event-related potential
disorder caused by nigrostriatal dopaminergic de®ciency. modi®cations induced by dopaminergic treatment (RuÊzÏicÏka
However, the presence of cognitive de®cits in PD also and El Massioui, 1993).
suggests disturbances on cortical and subcortico-cortical Our aim was to verify if patients in the early stages of PD
levels (Dubois et al., 1991). Indeed, as demonstrated by and before the introduction of l-DOPA treatment show any
Javoy-Agid and Agid (1980), the ascendant mesocortical electrophysiological indices of brain dysfunction. To
dopamine system may be affected as well. achieve this goal, we used both classical EEG spectral
Electroencephalography (EEG) is generally used to analysis and numerical methods based on non-linear
depict brain electrical activity re¯ecting the functional dynamics.
state of cortical layers and their subcortical driving struc- Spectral analysis and non-linear analysis are different
tures. A number of studies showed EEG abnormalities in PD ways of dealing with EEG apparent complexity. The former
patients, most commonly represented by a generalized slow- belongs to the theory of random process and, in that case,
ing of EEG activity (de Weerd et al., 1990; Soikkeli et al., the EEG signal is generally considered as a linear stochastic
1991; Neufeld et al., 1994). However, studies usually process. The latter belongs to dynamical systems theory
included patients in advanced stages of the disease where and, thus, EEG complexity is viewed as the result of non-
non-dopaminergic lesions supposedly contribute to general linear deterministic dynamics (possibly a `chaotic' process).
These two approaches gather each other when dealing with
the problem of signals characterized by a continuous spec-
trum (Berge et al., 1986) such as commonly observed in
* Corresponding author. Tel./fax: 1420-2-24-91-69-80.
E-mail address: [email protected] (E. RuÊzÏicÏka).
EEG. In fact, this continuous spectrum may be either related

1388-2457/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. CLINPH 2000614
PII: S13 88-2457(00)0051 2-5
 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45 39

to a quasi-periodic signal with a high (possibly in®nite) range 43±61 years) with no history or symptoms of neuro-
number of frequencies or to an aperiodic signal. If the signal logical or mental illness served as controls.
is indeed aperiodic, it may have been generated by a non- The scores for the Mini-Mental-Status examination were
linear deterministic system with a low number of degrees of within normal limits (28±30) in all patients and control
freedom. Therefore, the existence of low-dimensional non- subjects. Informed consent was obtained from all the
linear systems with complex aperiodic behaviors demon- subjects.
strates the possibility of a deterministic approach for
problems which were previously only statistically tractable. 2.2. EEG recordings
In the case of EEG, such a perspective may imply more
EEG was recorded by a multichannel apparatus Brainsco-
parsimonious models to account for EEG complexity.
pe q (M&I Ltd., Czech Republic) using 20 Ag-AgCl electro-
The application of non-linear analysis EEG considers the
des set on the scalp according to the 10±20 International
brain as a physical dynamical system (McKenna et al.,
Electrode Placement System (Jasper, 1958) (including the
1994). The putative importance of non-linear analysis has
Oz electrode but not the Fpz one) referenced to the linked
been previously emphasized (Jansen, 1991; Pritchard et al.,
mastoids. The EEG signal was digitized on 16 bits using a 1
1994; Micheloyannis et al., 1998) and reviewed (Pritchard
kHz frequency per channel and then bandpass ®ltered
and Duke, 1992; Elbert et al., 1994; Rey and Guillemant,
between 0.5 and 70 Hz. In order to eliminate ocular artifacts,
1997). The clinical importance of non-linear indices has
vertical and horizontal eye movements were simultaneously
been investigated in several neurological and psychiatric
recorded. All subjects were examined in the supine position
conditions such as Creutzfeld±Jacob disease (Babloyantz
during an eyes closed condition in a quiet and dimly
and Destexhe, 1988; Stam et al., 1997), Alzheimer's disease
shielded room. For each subject, 30 s of multichannel
(Pritchard et al., 1991; Pezard et al., 1998; Jelles et al.,
EEG signals free of signs of impaired wakefulness, ocular
1999), depression (Nandrino et al., 1994; Pezard et al.,
movements and other artifacts were selected for analysis.
1996b; Thomasson et al., 2000), schizophrenia (Roschke
et al., 1995) and epilepsy (Babloyantz and Destexhe, 2.3. Analysis of the EEG data
1986; Martinerie et al., 1998; Lehnertz et al., 2000a). For
books on the topic, see BasËar (1990), Duke and Pritchard 2.3.1. Segmentation of the EEC signals
(1991), Jansen and Brandt (1993), and Lehnertz et al. Each 30 s multichannel EEG signal was divided into 29
(2000b). segments of 1024 ms. The EEG segments were notch
Few studies have investigated the non-linear dynamics of ®ltered to eliminate 50 Hz artifacts.
EEG in PD (Stam et al., 1994, 1995). Their results mainly
showed speci®c patterns of dysfunction in dementia and PD 2.3.2. Linear analysis
on the basis of indices computed for the global brain elec- Linear analysis is based on the assumption that the linear
trical activity. Our study completes those previous ®ndings components (e.g. mean, autocorrelation, power spectrum)
(1) by computing linear and non-linear electrode-related are suf®cient to describe a signal. In particular, for spectral
indices and (2) by studying l-DOPA naive patients in the analysis, it is assumed that frequency bands are independent
®rst stages of the disease. and can be studied separately. Our linear analysis is based
on this standard assumption used in EEG quanti®cation.
The power spectrum of each multichannel EEG segment
2. Methods was computed on the basis of a Fast Fourier Transform of
the data. For each channel, the total power in each spectrum
2.1. Subjects was calculated by summing all components and the powers
in u-, a- and b-bands were calculated by summing compo-
After approval by the local ethics committee (General nents in bands from 4 to 8, 8 to 12 and 12 to 30 Hz, respec-
Faculty Hospital, Prague), we studied 9 patients with PD tively. The power in each band was divided by the total
(4 women, 5 men, mean age 54.3 years, range 48±63 years) power to provide u-, a- and b-band relative powers which
meeting the criteria for the diagnosis of idiopathic PD were used as our linear measurements.
(Ward and Gibb, 1990). The mean duration of PD was 3
years (range 1±6 years); 3 patients were in Hoehn and Yahr 2.3.3. Non-linear analysis
stage 1, 3 were in stage 1.5, and 3 were in stage 2. We
included only l-DOPA naive patients; 7 of them were 2.3.3.1. General principle. In non-linear analysis, a differ-
taking selegiline (5±10 mg), 6 were taking amantadine ent assumption is made, namely that some non-linear deter-
(200±300 mg), one was taking biperiden (2 mg), one was ministic structures are present in the data. Non-linear
taking benztropine (6 mg), one was taking alprazolam (0.25 indices are thus computed from the data. We choose here
mg), one was taking piracetam (1200 mg) and one patient two electrode-related measurements: slope asymmetry and
was without medication. local entropy (to be described below). Since linear proper-
Nine subjects (7 women, two men, mean age 51.3 years, ties may be responsible for the results of non-linear analysis,
40 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45

the non-linear measurements have to be statistically vali- curve is used to compute an entropy index (K) (Wales,
dated. Surrogate data which share the same linear character- 1991). The prediction curves obtained for each electrode
istics with the raw data are constructed (to be described permit the computation of local entropies (Ki for
below). Non-linear measurements are computed for those electrode i). A high entropy corresponds to a rapidly
surrogate data. A statistical procedure allows us to test decreasing predictability.
whether the measurements obtained from the raw data differ
from those obtained from the surrogate data. In that case,
non-linear structures are evidenced and, consequently, 2.3.3.4. Surrogate data testing. In order to ensure the
linear characteristics are not suf®cient for a reliable descrip- presence of non-linear structures in the brain dynamics,
tion of the data. On the contrary, if measurements obtained we proceeded to a surrogate data test. Namely,
from surrogate data do not differ from those obtained from multivariate linearly correlated noises that share the same
raw data, either no non-linear structure exists in the data or linear characteristics (in particular the power spectrum and
some drawbacks prevent its characterization. phase relations) with the observed signals were generated
(Prichard and Theiler, 1994). Non-linear indices (slope
2.3.3.2. Time series slope asymmetry. The concept of slope asymmetry and Ki) are computed for the surrogate signals
asymmetry is derived from the fact that time series and for the observed signals.
generated from linear processes (such as sine waves) In this study, two successive steps were followed.
appear statistically the same whether the data are viewed
as running forward or backward in time. By contrast, non- 1. Firstly, non-linearities were analyzed at a coarse level
linear systems generally have distinct rise and fall times. For using only one multivariate surrogate time series for
such systems, this measure captures the difference between each EEG segment and the presence of signi®cant non-
the rise times (upward part) and fall times (downward part) linearity was thus tested at the level of the entire set of
of the oscillations in a dynamical system. One of the EEG data.
simplest ways to measure slope asymmetry is by the 2. Secondly, a set of 39 multivariate surrogate data was
skewness (third statistical moment) of differences between generated for each EEG segment. This procedure allows
successive samples (Ehlers et al., 1998) one to test the presence of non-linearity at the ®ne level
P of individual EEG segment. This second step was
x 2 xi21 †3 performed using the non-linear indices for which signi®-
Slope asymmetry ˆ ÿP i
3=2
xi 2 xi21 †2 cant global non-linearities were observed. If the index
computed for the set of surrogate data signi®cantly
differs from that of the observed signals, it can be
2.3.3.3. Non-linear forecasting and local entropy. Local concluded that a non-linear process is involved in the
entropy measures the loss of predictability of the EEG generation of the observed EEG segment.
dynamics. It is computed using numerical methods based
on the principles of non-linear dynamics: trajectory In order to control possible false rejections of the linearity
reconstruction and non-linear forecasting (detailed assumption, we constructed arti®cial random multichannel
descriptions can be found in Pezard et al., 1994, 1996a). time series as follows (Rombouts et al., 1995)
1. The set of 1024 samples recorded over 20 electrodes is xk;t ˆ 0:9 xk;t21 1 1:05 xk;t21 2 xk;t22 † 1 e 1 0:015xk21;t †
considered as a set of 1024 vectors in a 20-dimensional
where xk,t is the amplitude value of channel k at time t (here
space. This ®rst step constitutes a multichannel
k ˆ 1, ¼, 20 and t ˆ 1, ¼, 1024) and e is independently
reconstruction procedure transforming the time series
distributed discrete noise. A set of 261 (9 £ 29) such linearly
into a multidimensional trajectory representing the
correlated noises was generated and divided into 9 sets of 29
evolution of brain activity (for a discussion of the
individual multichannel time series to mimic a group of 9
validity of reconstruction methods in the context of
subjects each associated with 29 EEG segments. These data
EEG, see Pezard et al., 1999; Pritchard, 1999).
were analyzed in the second step described above using the
2. The reconstructed trajectory was characterized using a
same procedure as the EEG data. They thus allow us to
non-linear forecasting method. It permits one to obtain
estimate the level of spurious rejections of the linearity
quanti®ers of the brain activity related to each recording
assumption.
site. Namely, a local linear model (Sugihara and May,
1990) is used to forecast the evolution of the vectors in 2.4. Statistical analysis of the EEG quanti®ers
the 20-dimensional space several time steps ahead. To
avoid correlation artifacts, vectors closer than 20 time Each patient was characterized by the averaged values
steps to the vector to be predicted were discarded from computed over the 29 EEG segments of u-, a- and b-band
the prediction model (Theiler, 1986). The decrease of the relative powers for the linear measurements and of slope
correlation coef®cient between the observed values and asymmetry, local entropy (Ki) and the number of non-linear
the predicted ones constitutes a prediction curve. This segments for the non-linear measurements.
 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45 41

Analyses of variance were used to test the effect of the

group factor (two levels, controls and patients) and of the
electrode factor (20 levels) for the linear measurements (u-,
a- and b-band relative powers). In the ®rst step of non-
linear analysis, an analysis of variance with 3 factors: type
of data (two levels, raw and surrogate), group (two levels,
controls and patients) and electrode (20 levels), was used to
characterize their effects on local entropy (Ki) and slope
asymmetry. In the case of the measurement where a signi®-
cant effect of the type of data was found, this index was used
in the second step of our non-linear analysis.
In the second step of the non-linear analysis, a ®rst stage
consists of testing the presence of false positive non-linear-
ity in our EEG data. For that purpose, we analyzed two 2 £ 2
contingency tables (Conover, 1999) comparing frequencies
of linearity assumption rejection for either controls versus
random data or PD patients versus random data. Since
signi®cant rejections of the linearity assumption were
observed (see Section 3), an analysis of variance was used
to test the effect of the group factor (two levels, controls and
patients) and of the electrode factor (20 levels) for the
number of non-linear segments obtained for EEG data.
Fig. 1. Relative powers of u-, a- and b-band of EEG rhythms for controls
and PD patients averaged over the 20 electrodes. Standard errors are repre-
sented with vertical lines. *P , 0:001.
3. Results
3.2. Non-linear analysis
3.1. Linear analysis
3.2.1. First step: coarse analysis
The results of spectral analysis are summed up in Table 1
for statistical analysis and in Fig. 1 for the difference The statistical results are summed up in Table 2. The
between groups. The decrease of the b-band relative slope asymmetry index detects non-linear structures (see
power in the patients is the only signi®cant difference Fig. 2), but does not depict any difference between groups
between controls and patients observed for linear measure- or electrodes. The local entropy index does not detect any
ments. The heterogeneity of brain activity is assessed by the non-linear structures but shows a signi®cant increase of
signi®cant effect of the electrode factor observed for the 3 entropy for patients (see Fig. 3) and an effect of the elec-
rhythms. No interaction between group and electrode
factors was observed, which does not permit one to search Table 2
for localized effects. Results of the analysis of variance for the non-linear measurements in the
coarse step (raw versus surrogate data)

Table 1 Effect d.f. F value P value

Results of the analysis of variance for the u-, a- and b-band relative powers
Local entropy
Effect d.f. F value P value Type 1 0.382 0.537
Group 1 9.597 0.002
u -band Electrode 19 4.779 ,0.001
Group 1 3.416 0.066 Type £ group 1 0.003 0.958
Electrode 19 6.115 ,0.001 Type £ electrode 19 0.004 1.000
Group £ electrode 19 0.273 0.999 Group £ electrode 19 0.359 0.995
Type £ group £ electrode 19 0.004 1.000
a -band
Group 1 2.444 0.119 Slope asymmetry
Electrode 19 1.642 0.045 Type 1 11.999 ,0.001
Group £ electrode 19 0.067 1.000 Group 1 1.015 0.314
Electrode 19 0.703 0.817
b -band Type £ group 1 1.419 0.234
Group 1 13.720 ,0.001 Type £ electrode 19 0.732 0.787
Electrode 19 2.812 ,0.001 Group £ electrode 19 0.594 0.912
Group £ electrode 19 0.154 0.999 Type £ group £ electrode 19 0.805 0.703
42 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45

Table 3
Results of the analysis of variance for the number of EEG segments where
signi®cant non-linear structures were found in the ®ne step of the non-linear
analysis

Effect d.f. F value P value

Group 1 5.641 0.018

Electrode 19 1.269 0.202
Group £ electrode 19 0.453 0.978

cantly higher than the level of spurious rejections for both

groups (T12 ˆ 107:5, P , 0:001 for controls; T12 ˆ 204:3,
P , 0:001 for PD patients). Signi®cant non-linear struc-
tures were thus detected in our EEG data at the level of
individual EEG segments. The effect of the notch ®lter on
the frequency of linearity assumption rejection was tested
Fig. 2. Difference between raw and surrogate data shown by slope asym- using the same procedure in the case of the control group.
metry (S. A. £ 10 24) averaged over the 20 electrodes for both groups.
No signi®cant difference between notch-®ltered and
Standard errors are represented with vertical lines. *P , 0:001.
unnotch-®ltered data was observed (T12 ˆ 0:88, P . 0:75).
The effect of the notch-®lter on the non-linearity level can
trode factor. Nevertheless, the absence of signi®cant inter- thus be dismissed.
action does not permit one to test for localized effects. The statistical results obtained for the comparison
between EEG data in controls and PD patients are summed
up in Table 3. The number of EEG segments where signi®-
3.2.2. Second step: ®ne analysis
cant non-linear structure was found is higher in PD patients
On the basis of the results obtained in the ®rst step, the
than in controls (see Fig. 4). Nevertheless, non-linear struc-
®ne analysis was performed using the slope asymmetry
ture does not vary with electrode.
index.
The frequencies of the linearity assumption rejection
were compared between EEG data (controls and PD
patients) and arti®cial multichannel random data. The 4. Discussion
frequency of rejection of the linearity assumption is signi®-
PD is known as a predominantly motor disorder produced
by dopaminergic de®ciency in the basal ganglia. However,
non-motor impairment including cognitive dysfunction has
frequently been observed in the early stages of PD (Cooper
et al., 1991). Electrophysiological methods can help to
investigate various non-motor aspects of the disease inde-
pendently from basic motor impairment; for example,
event-related potentials allow one to study the speed of
cognitive processing in speci®c tasks (see RuÊzÏicÏka and El

Fig. 3. Local entropy (kKl in s 21) for controls and PD patients averaged over
the 20 electrodes. Standard errors are represented with vertical lines. Fig. 4. Number of non-linear EEG-segments (# Seg.) averaged over the 20
*P , 0:01. electrodes. Standard errors are represented with vertical lines. *P , 0:05.
 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45 43

Massioui, 1993, for review). Furthermore, EEG analysis EEG data in PD patients compared to controls. We checked
(linear and non-linear methods) depicts more global indices the validity of these results for group comparisons. Namely,
of brain function that can re¯ect disturbed subcortico-corti- we investigated possible spurious detection of non-linear
cal mechanisms in patients with advanced PD and/or structure and show that the number of EEG segments
dementia (de Weerd et al., 1990; Stam et al., 1994, 1995; where non-linear structures were detected is above the
Pezard et al., 1998). The goal of our study was to search for level of false detection. Thus, variations in this number of
such electrophysiological indices in less advanced cases of EEG segments can be considered as signi®cant for the
PD. With that in mind, we examined both linear and non- comparison between controls and PD patients, although
linear EEG characteristics in a group of non-demented the level of non-linear segments remains low (between 1
patients in the early stages of PD without l-DOPA medica- and 2%, i.e. similar to the results observed for alpha rhythm
tion. in healthy volunteers by Stam et al. (1999)). Nevertheless,
With regard to the linear EEG indices, we observed a our ®ndings are contrary to those obtained in studies of
decrease in the relative power of the b-band in PD patients Alzheimer's disease (Jelles et al., 1999) or of ethanol intake
compared to controls. The a- and u-band relative powers (Ehlers et al., 1998) where a reduction of non-linear struc-
were not signi®cantly different from controls. On the tures in EEG data compared to controls was reported. We
contrary, in previous studies in more advanced stages of assume that the effect of ethanol and Alzheimer's disease on
PD accompanied with cognitive decline, generalized slow- cerebral dynamics is different than that observed in the case
ing of EEG was observed with an increase in d- and u-band of PD.
(de Weerd et al., 1990; Soikkeli et al., 1991). Therefore, we A number of previous PD studies were done in chroni-
may hypothesize that the isolated diminution of b-band cally treated patients in whom the effects of disease chroni-
power represents the earliest linear EEG marker of initial city and possible dementia might in¯uence the results. In an
cortical or subcortico-cortical dysfunction in PD. attempt to avoid such a confusion, we included in the
The linear indices are not suf®cient to fully describe the present study only non-demented patients in early stages
complexity of brain activity since the comparison between of PD without l-DOPA treatment. However, since non-
slope asymmetry of raw EEG signals and surrogate data linear EEG measures may be sensitive to medication effects
showed the presence of signi®cant non-linear structures in (Wackermann et al., 1993; Pezard et al., 1998), possible
the EEG of both PD patients and controls. in¯uences of other drugs have to be considered. Most of
Employing our two non-linear indices (slope asymmetry our patients were receiving a combination of drugs with
and entropy) to study the difference between PD patients putative neuroprotective and/or weak symptomatic effects
and controls, we observed that the EEG of PD patients is (selegiline, amantadine). Nevertheless, these drugs do not
characterized by a higher presence of non-linear EEG appear to signi®cantly in¯uence either cognition or electro-
segments and by a higher entropy. High entropy means physiological indices in early stages of PD (Gelenberg et al.,
that the quality of prediction of EEG dynamics decreases 1989; Dalrymple-Alford et al., 1995; Ziemann et al., 1997).
more rapidly in PD patients than in controls. Thus, the local Our results are thus more probably related to disease effects
linear model used to forecast the time-evolution of EEG is than to medication effects.
better in the case of controls than in PD patients. In other In conclusion, the reduced power of b-band and the
words, the 20-dimensional trajectory of PD patients increase of entropy and the number of non-linear EEG
contains more non-linear structures than that of the controls. segments may represent the early signs of subcortico-corti-
This result suggests that with the trajectory reconstruction, cal dysfunction in PD. Studies following the disease
less non-linear structure is lost in PD patients than in progression and treatment effects might be able to demon-
controls (Hegger et al., 1998). Since it has been shown strate if it is namely the impairment of the dopaminergic
that the amount of detected non-linearity decreases with system in PD that causes this modi®cation of brain
the augmentation of dimension (Lachaux et al., 1997), we dynamics.
can hypothesize that PD would be associated with brain
dynamics of a lower dimension than that of controls as
previously reported (Stam et al., 1994). Acknowledgements
The modi®cations of brain dynamics in early stages of PD
were observed only at a global level without any signi®cant The study was partly supported by Charles University
localization for both linear and non-linear indices. Never- (project CEZ J13/98 111100001) and by the Scienti®c
theless, linear indices (u-, a- and b-band relative powers) Service of the French Embassy in Prague. The authors are
and entropy index depicted a signi®cant effect of the elec- thankful to Dr J. Roth and Dr P. Mecir who kindly referred
trode factor whereas slope asymmetry does not. Spectral two patients and to O. KucÏerova for her technical assistance
indices and entropy are thus more speci®c than slope asym- with EEG recordings. L.P. gratefully acknowledges fruitful
metry in dealing with spatial heterogeneity of brain func- discussions with J. Martinerie and I. Rivals. The comments
tioning. of anonymous reviewers on a previous version of the manu-
An increase in non-linear structure was observed here for script allowed us to improve this article.
44 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45

References McKenna TM, McMullen TA, Shlesinger MF. The brain as a physical
dynamical system. Neuroscience 1994;60:587±605.
Agid Y, Javoy-Agid F, Ruberg M. Biochemistry of neurotransmitters in Micheloyannis S, Flitzanis N, Papanikolaou E, Bourkas M, Terzakis D,
Parkinson's disease. In: Marsden CD, Fahn S, editors. Movement disor- Arvanitis S, Stam CJ. Usefulness of non-linear EEG analysis. Acta
Neurol Scand 1998;97:13±19.
ders, vol. 2. London: Butterworths, 1987. pp. 166±230.
Nandrino J-L, Pezard L, Martinerie J, El Massioui F, Renault B, Jouvent R,
Babloyantz A, Destexhe A. Low-dimensional chaos in an instance of
Allilaire J-F, WidloÈcher D. Decrease of complexity in EEG as a symp-
epilepsy. Proc Natl Acad Sci USA 1986;83:3513±3517.
tom of depression. NeuroReport 1994;5:528±530.
Babloyantz A, Destexhe A. The Creutzfeld-Jacob disease in the hierarchy
Neufeld MY, Blumen S, Aitkin I, Parmet Y, Korczyn AD. EEG frequency
of chaotic attractors. In: Markus M, MuÈller S, Nicolis G, editors. From
analysis in demented and nondemented parkinsonian patients. Demen-
chemical to biological organization, Springer series in synergetics, 39.
tia 1994;5:23±28.
Heidelberg: Springer-Verlag, 1988. pp. 307±316.
Pezard L, Martinerie J, Breton F, Bourzeix J-C, Renault B. Nonlinear
BasËar E, editor. Chaos in brain function New York: Springer-Verlag, 1990.
forecasting measurements of multichannel EEG dynamics. Electroen-
Berge P, Pomeau Y, Vidal C. Order within chaos, New York: Wiley, 1986.
ceph clin Neurophysiol 1994;91:383±391.
Conover WJ. Practical nonparametric statistics, 3rd ed. New York: Wiley,
Pezard L, Martinerie J, MuÈller J, Varela FJ, Renault B. Entropy quanti®ca-
1999.
tion of human brain spatio-temporal dynamics. Physica D
Cooper JA, Sagar HJ, Jordan N, Harvey NS, Sullivan EV. Cognitive
1996a;96:344±354.
impairment in early, untreated Parkinson's disease and its relationship
Pezard L, Nandrino J-L, Renault B, El Massioui F, Allilaire J-F, MuÈller-
to motor disability. Brain 1991;114:2095±2122.
Gerking J, Varela FJ, Martinerie J. Depression as a dynamical disease.
Dalrymple-Alford JC, Jamieson CF, Donaldson IM. Effects of selegiline
Biol Psychiatry 1996b;39:991±999.
(deprenyl) on cognition in early Parkinson's disease. Clin Neurophar-
Pezard L, Martinerie J, Varela FJ, Bouchet F, Derouesne C, Renault B.
macol 1995;18:348±359.
Brain entropy maps quantify drug dosage on Alzheimer's disease.
de Weerd AW, Perquin WVM, Jonkman EJ. Role of the EEG in the predic-
Neurosci Lett 1998;253:5±8.
tion of dementia in Parkinson's disease. Dementia 1990;1:115±118.
Pezard L, Lachaux J-P, Thomasson N, Martinerie J, et al. Why bother to
Dubois B, Boller F, Pillon B, Agid Y. Cognitive de®cits in Parkinson's
spatially embed EEG? Comments on Pritchard et al. Psychophysiology
disease. In: Boller F, Grafman J, editors. Handbook of neuropsychol-
33 (1996) 362-368. Psychophysiology 1999;36:527±531.
ogy, 5. Amsterdam: Elsevier, 1991. pp. 195±240.
Pillon B, Dubois B, Cusimano G, Bonnet AM, Lhermitte F, Agid Y. Does
Duke DW, Pritchard WS, editors. Measuring chaos in the human brain
cognitive impairment in Parkinson's disease result from non-dopami-
Singapore: World Scienti®c, 1991. nergic lesions?. J Neurol Neurosurg Psychiatry 1989;52:201±206.
Ehlers CL, Havstad J, Prichard D, Theiler J. Low doses of ethanol reduce
Prichard D, Theiler J. Generating surrogate data for time series with several
evidence for nonlinear structure in brain activity. J Neurosci simultaneous measured variables. Phys Rev Lett 1994;73:951±954.
1998;18:7474±7486. Pritchard WS, Duke DW. Measuring chaos in the brain: a tutorial review of
Elbert T, Ray WJ, Kowalik ZJ, Skinner JE, Graf KE, Birbaumer N. Chaos nonlinear dynamical EEG analysis. Int J Neurosci 1992;67:31±80.
and physiology: deterministic chaos in excitable cell assemblies. Pritchard WS, Duke DW, Coburn KL. Altered EEG dynamical responsivity
Physiol Rev 1994;74:1±47. associated with normal aging and probable Alzheimer's disease.
Gelenberg AJ, van Putten T, Lavori PW, Wojcik JD, Falk WE, Marder S, Dementia 1991;2:102±105.
Galvin-Nadeau M, Spring B, Mohs RC, Brotman A. Anticholinergic Pritchard WS, Duke DW, Coburn KL, Moore NC, Tucker KA, Jann MW,
effects on memory: benztropine versus amantadine. J Clin Psychophar- Hostetler RM. EEG-based, neural-net predictive classi®cation of
macol 1989;9:180±185. Alzheimer's disease versus control subjects is augmented by non-linear
Hegger R, Kantz H, Olbrich E. Problems in the reconstruction of high- EEG measures. Electroenceph clin Neurophysiol 1994;91:118±130.
dimensional deterministic dynamics from time series. In: Kantz H, Pritchard WS, et al. On the validity of spatial embedding: a reply to Pezard
Kurth J, Mayer-Kress G, editors. Nonlinear analysis of physiological et al. Psychophysiology 1999;36:532±535.
data, Berlin: Springer-Verlag, 1998. pp. 23±47. Rey M, Guillemant P. Apport des matheÂmatiques non-lineÂaires (theÂorie du
Jansen BH. Quantitative analysis of electroencephalograms: is there chaos chaos) aÁ l'analyse de l'EEG. Neurophysiol Clin 1997;27:406±428.
in the future?. Int J Biomed Comput 1991;27:95±123. Rombouts SARB, Keunen RWN, Stam CJ. Investigation of nonlinear struc-
Jansen BH, Brandt ME, editors. Nonlinear analysis of the EEG Singapore: ture in multichannel EEG. Phys Lett A 1995;202:352±358.
World Scienti®c, 1993. Roschke J, Fell J, Beckmann P. Nonlinear analysis of sleep EEG data in
Jasper EA. The ten twenty electrode system of the international federation. schizophrenia: calculation of the principal Lyapunov exponent.
Electroenceph clin Neurophysiol 1958;10:371±375. Psychiatry Res 1995;56:257±269.
Javoy-Agid F, Agid Y. Is the mesocortical dopaminergic system involved RuÊzÏicÏka E, El Massioui F. Event-related potentials in Parkinson's disease: a
in Parkinson disease?. Neurology 1980;30:1326±1330. review. Behav Neurol 1993;6:15±26.
Jelles B, van Birgelen JH, Slaets JP, Hekster RE, Jonkman EJ, Stam CJ. Soikkeli R, Partanen J, Soininen H, Paakkonen A, Riekkinen Sr. P. Slowing
Decrease of non-linear structure in the EEG of Alzheimer patients of EEG in Parkinson's disease. Electroenceph clin Neurophysiol
compared to healthy controls. Clin Neurophysiol 1999;110:1159±1167. 1991;79:159±165.
Lachaux J-P, Pezard L, Garnero L, Pelte C, Renault B, Varela FJ, Martin- Stam CJ, Woerkom TC, Keunen RW. Non-linear analysis of the electro-
erie J. The spatial extension of brain activity fools the single-channel encephalogram in Creutzfeld-Jacob disease. Biol Cybern 1997;77:247±
reconstruction of EEG dynamics. Hum Brain Map 1997;5:26±47. 256.
Lehnertz K, Andrzejak RG, Arnhold J, Widman G, Burr W, David P, Elger Stam CJ, Pijn JPM, Suffczynski P, Lopes da Silva FH. Dynamics of the
CE. Possible clinical and research applications of nonlinear EEG analy- human alpha rhythm: evidence for non-linearity? Clin Neurophysiol
sis in humans. In: Lehnertz K, Arnhold J, Grassberger P, Elger CE, 1999;110:1801±1813.
editors. Chaos in brain? Singapore: World Scienti®c, 2000a. pp. 134± Stam KJ, Tavy DL, Jelles B, Achtereekte HAM, Slaets JPJ, Keunen RWM.
155. Non-linear dynamical analysis of multichannel EEG: clinical applica-
Lehnertz K, Arnhold J, Grassberger P, Elger CE, editors. Chaos in brain? tions in dementia and Parkinson's disease. Brain Topogr 1994;7:141±
Singapore: World Scienti®c, 2000b. 450.
Martinerie J, Adam C, Le van Quyen M, Baulac M, Renault B, Varela FJ. Stam KJ, Jelles B, Achtereekte HAM, Rombouts SARB, Slaets JPJ, Keunen
Epileptic crisis can be anticipated by non-linear analysis. Nat Med RWM. Investigation of EEG non-linearity in dementia and Parkinson's
1998;4:1173±1176. disease. Electroenceph clin Neurophysiol 1995;95:309±317.
 L. Pezard et al. / Clinical Neurophysiology 112 (2001) 38±45 45

Sugihara G, May RM. Nonlinear forecasting as a way of distinguishing chaos functional state after a single dose of nootropic drug. Electroenceph
from measurement error in time series. Nature 1990;344:734±741. clin Neurophysiol 1993;86:193±198.
Theiler J. Spurious dimension from correlation algorithms applied to Wales DJ. Calculating the rate of loss of information from chaotic time
limited time-series data. Phys Rev A 1986;34:2427±2432. series by forecasting. Nature 1991;350:485±488.
Thomasson N, Pezard L, Allilaire J-F, Renault B, Martinerie J. Nonlinear Ward CD, Gibb WR. Research diagnostic criteria for Parkinson's disease.
EEG changes associated with clinical improvement in depressed Adv Neurol 1990;53:245±249.
patients. Nonlinear Dynam Psychol Life Sci 2000;4:203±218. Ziemann U, Tergau F, Bruns D, Baudewig J, Paulus W. Changes in human
Wackermann J, Lehmann D, DvorÏak I, Michel CM. Global dimensional motor cortex excitability induced by dopaminergic and anti-dopaminer-
complexity of multi-channel EEG indicates change of human brain gic drugs. Electroenceph clin Neurophysiol 1997;105:430±437.