Page 1

Fast track report

Quantitative EEG in early Alzheimer's disease patients — Power spectrum

and complexity features

Balázs Cziglera,⁎, Dóra Csikósa, Zoltán Hidasib, Zsófia Anna Gaála, Éva Csibrib,

Éva Kissc, Pál Salaczb, Márk Molnára

aInstitute for Psychology, Hungarian Academy of Sciences, Budapest, Hungary

bDepartment of Psychiatry, Semmelweis University Medical School, Budapest, Hungary

cNational Institute of Psychiatry and Neurology, Budapest, Hungary

Received 3 October 2007; received in revised form 13 November 2007; accepted 14 November 2007

Available online 22 November 2007

Abstract

The goal of this study was to investigate the EEG signs of early stage Alzheimer's disease (AD) by conventional analyses and by methods

quantifying linear and nonlinear EEG-complexity. The EEG was recorded in 12 mild AD patients and in an age-matched healthy control

group (24 subjects) in both eyes open and eyes closed conditions. Frequency spectra, Omega-complexity and Synchronization likelihood were

calculated on the data. In the patients a significant decrease of the relative alpha and increase of the theta power were found. Remarkably

increased Omega-complexity and lower Synchronization likelihood were observed in AD in the 0.5–25 Hz frequency ranges. It is concluded

that both spectral- and EEG-complexity changes can be found already in the early stage of AD in a wide frequency range. Application

of conventional EEG analysis methods in combination with quantification of EEG-complexity may improve the chances of early diagnosis

of AD.

© 2007 Elsevier B.V. All rights reserved.

Keywords: Alzheimer's disease; Quantitative EEG; Nonlinear dynamics; Omega-complexity; Synchronization likelihood

1. Introduction

7.6 million people were reported suffering from dementia

in Europe in 2000 (Blennow et al., 2006) which makes it one

of the most frequent diseases amongst the elderly. Dementia

is caused by Alzheimer's disease (AD) in 40–60% of the

cases, therefore it is the most prevalent of neurodegenerative

diseases (Alloul et al., 1998). These data underscore the im-

portance of early diagnosis and the need to develop ap-

propriate methods for monitoring therapy and the course of

the disease.

Different phases of the disease can be defined based on the

progressive development of neurofibrillary tangles and amyloid

plaques (Braak and Braak, 1995; Braak and Del Tredici, 2004).

Impairment of the long corticocortical tracts (Wurtman et al.,

1996) and synaptic degeneration (Vellas and Fitten, 2000) were

also documented.

The rate of correctly identified AD-cases by electrophysio-

logical methods varies within a wide range, between 29% and

42% in the early and between 60% and 80% in the later stage

(Stam et al., 1996). The most frequently observed spectral

change is the “slowing” of the EEG, which supposedly cor-

relates with the degree of cognitive decline (Bennys et al.,

2001). Increase of theta and decrease of beta power is expected

to appear in the early stages, followed by a decrease of alpha

frequency (Robinson et al., 1994). In severely demented AD

patients the decrease of alpha activity is accompanied by an

increase of delta power (Coben et al., 1985; Dringenberg,

2000).

Available online at www.sciencedirect.com

International Journal of Psychophysiology 68 (2008) 75–80

www.elsevier.com/locate/ijpsycho

⁎Corresponding author. Institute for Psychology of the Hungarian Academy

of Sciences, Budapest POB 398, H-1394 Hungary. Tel.: +36 1 354 2290;

fax: +36 1 354 2416.

E-mail address: bazsoczi@yahoo.com (B. Czigler).

0167-8760/$ - see front matter © 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.ijpsycho.2007.11.002

Page 2

The study of complex systems such as the brain, showing

both linear and nonlinear features requires appropriate techni-

ques for the analysis of its functional properties, made possible

only by recent advances in the mathematics of nonlinear

dynamical systems (Stam, 2005). It is to be expected that since

interneuronal connections are impaired in AD these changes can

be revealed by “complexity analyses” of the EEG (Pijnenburg

et al., 2004). In this context, “dynamical complexity” is related

to the level of interactions (synchrony) between different parts

of a functional network (Stam, 2005). These methods have

provided useful information in AD (for reviews, see Jeong,

2004; Stam 2005). Most of the studies in this field were con-

cerned with global characteristics of the EEG and did not

investigate possible changes of individual frequency bands

(from theta to gamma), although their different functional roles

have become well known (Klimesch, 1999; Klimesch et al.,

2007 and others).

The aim of the present study was to investigate linear and

nonlinear EEG-complexity in various EEG-frequency bands to

obtainadditionalinformationinearlystageADpatientscompared

tothetraditionalspectralanalysis.Synchronizationchangesofthe

fast and slow EEG-frequencies correspond to various cognitive

processes, depending on the efficiency of short-, and long-term

connectivities (Buzsaki and Draguhn, 2004). Thus, it seemed

reasonable to hypothesize that already in the early AD patients

thesenewmethodswillrevealdecreasedEEG-synchronizationin

various frequency bands corresponding to different aspects of

cognitive decline.

2. Methods

2.1. Participants

12 mild AD patients (10 females, mean age 72.34 years, SD:

5.9 years) were studied. The diagnosis of probable Alzheimer's

disease was based on the criteria of NINCDS–ADRDA

(McKhann et al., 1984). All patients were subjected to internal

medical, neurological and psychiatric examination and were

evaluated on diagnostic (Hachinski et al., 1974; Hamilton 1960)

andstaging(MMSE,Folsteinetal.,1983) scales.Allpatientshad

CTor MR scans.

Inclusion criteria for the AD group were: probable AD by the

criteria mentioned above, MMSE score between 20 and 24, age

above 50 years, unaltered medication in the previous 3 months.

Exclusion criteria were: cause of dementia other than AD ac-

cording to CT, MRI or laboratory findings, clinical signs of

vascular dementia or Hachinski score above 4, severe depres-

sion (Hamilton Depression score beyond 15). Patients with

other neurological diseases, impairment of hearing, taking

medication that influences the EEG (e.g. neuroleptic drugs,

benzodiazepines, cholinesterase inhibitors) were also excluded

from the study. The patients were directed to our laboratory

from the Semmelweis University's Teaching Hospital of

Psychiatry and Psychotherapy, Budapest, and from the National

Institute of Psychiatry and Neurology, Budapest.

The control group consisted of 24 healthy elderly subjects

(16 females, MMSE 29–30, mean age 72.5 years, SD:

5.62 years), who were compensated for their participation. The

study was allowed by the Ethics Committee of the Institute of

Psychology of the Hungarian Academy of Sciences and also by

that of the Semmelweis University's Teaching Hospital of

Psychiatry and Psychotherapy.Allsubjectsgave their consent in

the study in writing.

2.2. Data acquisition

2 min EEG-recordings were performed both in “eyes

closed” and “eyes open” conditions, (21 electrodes positioned

according to the international 10–20 system, nose as reference,

linked earlobes as ground). Vertical and horizontal eye move-

ments were recorded by electrodes placed above and beneath

the right eye, and in the lateral canthi. NuAmps system was

used for recording (A/D frequency: 1000 Hz, bandpass filter:

0.5–45 Hz, 10% Cosine window, FIR). The EEG was seg-

mented into 2048 ms epochs. Sweeps exceeding +/−70 µV

were excluded by automatic artifact rejection, which was

followed by visual artifact screening. In the case of two control

subjects andone patient the total lengthof the artifact-free EEG

did not exceed 30 s and so these subjects were omitted from

the analyses. The average number of artifact-free epochs per

person was 28.4.

2.3. Data analyses

Spectral and complexity analyses were performed in the

following frequency bands: delta: 0.5–4 Hz, theta: 4–8 Hz,

alpha1: 8–11 Hz, alpha2: 11–14 Hz, beta1: 14–25 Hz, beta2:

25–35 Hz. Calculation of relative spectra for the frequency

ranges described above was performed by Neuroscan 4.3 and

MATLAB 7.1 softwares. Statistical analyses were carried out

for the whole scalp (as the average of the spectral values

measured on all electrodes), for the anterior (Fp1, Fp2, F3,

F4, F7, F8, Fz) and posterior (P3, P4, O1, O2, T5, T6, Pz)

areas, for the right side (Fp2, F4, C4, P4, O2, F8, T4, T6) and

for the left side (Fp1, F3, C3, P3, O1, F7, T3, T5). Three-

way ANOVAs were calculated in Group (AD vs. control)×

Condition (eyes open vs. closed)×Localization (anterior vs.

posterior) and in Group×Condition×Side (right vs. left)

designs.

For complexity analyses the DigEEG software was used for

computing Synchronization likelihood and Omega-complexity

on the same data (C.J. Stam, VU University Medical Center,

Amsterdam, by permission). A detailed description of these

methods can be found in papers by Stam and van Dijk (2002)

and Wackermann (1999). The basic principles of the methods

are summarized below.

Synchronization likelihood is based on comparison of

vectors representing the state of the system reconstructed in

state space by time-delay embedding. It is a measure of the

linear and nonlinear interdependency (strength of coupling)

between a time series and one or more other time series (e.g.

EEG-channels). In case of total synchrony the value of

Synchronization likelihood is 1 while for completely indepen-

dent systems it equals 0.

76 B. Czigler et al. / International Journal of Psychophysiology 68 (2008) 75–80

Page 3

Omega-complexity is obtained by spatial principle compo-

nent analysis on the EEG-channel data where the entropy of the

normalized eigenvalues of the components is defined as the

logarithm of Omega. If one component can describe all the time

series, it means that a single source (“mechanism”) would be

responsible for the dynamics of the EEG. Low and high Omega

values correspond to high and low levels of synchronization,

respectively.

For the data shown in the present study Synchronization

likelihood and Omega-complexity were calculated for the

whole scalp, taking all electrodes into account.EEG-complexity

parameters were analyzed by computing Group×Condition

analyses of variance.

3. Results

3.1. Frequency spectral analysis

In the theta band significant between-group differences were

observed. Relative theta power was higher in the AD group in

both eyes closed F(1,29)1=24.81; pb0.001) and eyes open

(F=6.21; pb0.02) conditions for the whole scalp (means of all

electrodes); for the left side in eyes closed (F=21.06, pb0.001)

and eyes open (F=5.63; pb0.03) conditions, for the right side

in eyes closed (F=18.14; pb0.001) and eyes open (F=9.85;

pb0.01) conditions, for the anterior (eyes closed: F=6.32;

pb0.02; eyes open: F=4.8; pb0.05) and the posterior (eyes

closed: F=11.62; pb0.001; eyes open: F=7.56; pb0.01)

areas.

Significant and marginally significant differences were also

revealed in the alpha2 band, where a decrease in relative power

was seen in the AD group compared to the control group for

the whole scalp (eyes closed: F=5.1; pb0.05 and eyes open:

F=4.1; pb0.05), for the left side (eyes closed: F=6.14;

pb0.02 and eyes open: F=4.81; pb0.05), for the right side

(eyes closed: F=4.13; pb0.05 and eyes open: F=3.3;

pb0.08), for the anterior (eyes closed: F=4.52; pb0.05; not

significant in the eyes open condition), and for the posterior

(eyes closed: F=3.2; pb0.09 and eyes open: F=4.13; pb0.05)

regions. Group averages of alpha2 and theta bands are shown in

Fig. 1.

In both groups opening the eyes resulted in a significant

decrease in the alpha1 (F=32.141; pb0.0001) and an increase

in the beta1 (F=22.091; pb0.0001) and beta2 (F=13.091;

pb0.001) bands. No significant Group×Condition (eyes closed

vs. open) interaction was obtained.

3.2. Complexity analyses

Synchronization likelihood was significantly lower in AD in

all frequency bands in both eyes closed and eyes open con-

ditions except for the beta2 band where this difference was not

significant in either condition. The significant Synchronization

likelihood differences found in the various frequency bands and

recording conditions were the following: delta (eyes closed:

F=4.68; pb0.05 and eyes open: F=6.75; pb0.02); theta (eyes

closed: F=6.75; pb0.02 and eyes open: F=11.53; pb0.01);

alpha1 (eyes open: F=8.4; pb0.01); alpha2 (eyes closed

condition F=5.48; pb0.05); beta1 band (eyes closed condition

F=4.3; pb0.05). A marginally significant difference was also

found in the alpha1 band in the eyes closed condition (F=3.32;

pb0.08), (see Fig. 2 for group averages in each frequency band

and condition).

No significant Group×Condition interaction or Condition

main effect was observed for Synchronization likelihood.

Omega-complexity was significantly higher in the AD

group for the delta band in the eyes open (F=4.43; pb0.05),

and for the theta band in both conditions (eyes closed:

F=4.73; pb0.05 and eyes open: F=6.86; pb0.02), for the

alpha1 band in the eyes open condition (F=6.9; pb0.02), for

the alpha2 band in both conditions (F=7.18; pb0.02 and

F=6.34; pb0.02, respectively), and for the beta1 range for

the eyes closed condition (F=8.42; pb0.01). A marginally

significant difference was found in the eyes closed condition

for the delta frequency range (F=3.52; pb0.07) and the

alpha1 range (F=3.67; pb0.07). Group averages for each

band and condition are shown in Fig. 2. As a result of open-

ing the eyes Omega increased considerably in both groups

1If not stated otherwise the degrees of freedom were (1,29).

Fig. 1. Relative power (%) of theta and alpha2 bands in eyes open and eyes

closed conditions in the AD and control groups. Vertical bars denote 95%

confidence intervals.

77 B. Czigler et al. / International Journal of Psychophysiology 68 (2008) 75–80

Page 4

(F=9.6096; pb.005). No significant Group×Condition effect

was found.

4. Discussion

Our findings support earlier data showing spectral “slowing”

of EEG in AD (Coben et al., 1985; Brenner, 1999; van der Hiele

etal.,2007).Thelossofcholinergicneuronsinthebasalforebrain

projecting to the hippocampus and the neocortex may play an

important role in this process (Dringenberg, 2000) It is to be

noted, however, that the malfunction of other neurotransmitter

systems such as the monoaminergic systems is also involved in

AD (Wong et al., 2006).

According to our findings the amount of relative theta band

increased and that of the fast alpha range decreased in AD. This

effect could be seen both in the eyes closed and eyes open

conditions, in both hemispheres and also for both the anterior

and posterior areas. It is to be emphasized that in the present

study these changes were detected in the very early stage of

dementia.

Former studies foundlower dimensional complexityinADby

the correlation dimension method (Jeong, 2004) and also by

calculating the Lempel–Ziv complexity at a few electrode sites

(Abasolo etal.,2006).Inthese formerstudies differentfrequency

bands were not analyzed separately. More recently, a decrease of

synchronization in the beta frequency band was observed by the

SynchronizationlikelihoodmethodinmildAD(Pijnenburgetal.,

2004).

Inourstudy,Omega-complexitywasfoundtobehigher,while

Synchronization likelihood was lower in the patient group in all

frequency bands both in eyes closed and eyes open conditions

except for the fastest (beta2) frequency band. The increase of

Omega-complexity and the decrease of Synchronization like-

lihood correspond to an increased number of independent neural

Fig. 2. Synchronization likelihood and Omega-complexity in the two groups in the six frequency bands, in eyes open and eyes closed conditions (averages).

78 B. Czigler et al. / International Journal of Psychophysiology 68 (2008) 75–80

Page 5

assemblies and thus probably indicate the disintegration of

different parts of the nervous system, including different regions

of the neocortex. This is to be expected, given the neuropatho-

logical changes characterizing AD, which compromise the

structuralbasisofinterneuralcooperationandintegrativecapacity

of the brain, indispensable for cognitive activity. The reduced

coherence found in AD in the alpha and beta bands (Locatelli

et al., 1998; Jeong et al., 2001; Adler et al., 2003; Hogan et al.,

2003,thelatterin“mildAD”)alsosupportsthisview.Themodest

complexity-changes in the beta2 band may indicate that very

short-range neuronalconnectivityisrelativelysparedinthisearly

stage of the disease.

While the spectral measures showed changes only in the theta

andalpha2 bands in the AD group,results of the EEG-complexity

analyses verified decreased synchronization in all frequency

bands,exceptforbeta2.Sinceouranalysisdoesnotfullycoverthe

“classical” gamma band, in which lower synchronization was

found in AD by Stam and van Dijk (2002), based on the present

findingsitisnotpossibletodrawinferencesifcognitiveperceptual

processesrepresentedbythisrangeareinvolved.However,thefact

that consistent changes indicative of lower degree of synchroniza-

tion were observed in all other frequency bands suggest that

already in early AD the capacity for both short-, and long-range

connectivity are reduced. Such reduction may provide the basis

for the compromise of various types of cognitive processes

(like attention, memory, etc.). Our finding, that a wide range of

frequency bands are involved in AD supports that of Koenig et al.

(2005)whoapplieda new method(GlobalField Synchronization)

forthecalculationofsynchronizationalthoughourstudyisunique

inthatonlyearlyADpatientswereinvestigatedandbothlinearand

nonlinear methods were applied in the analyses.

Although the number of subjects in the present study is rela-

tivelylow,the statisticallysignificant findings mayserve as a basis

for further studies in which the severity of AD is correlated with

spectral and complexity analyses. According to the present ob-

servations, both Omega-complexity and Synchronization like-

lihoodprovedtobesensitivemethodsfortheassessmentofpatients

with Alzheimer's disease complementing spectral measures. It

remains to be seen if the results provided by these methods will

prove to be sensitive indicators of the progression of pathological

processes such as those seen in AD and, correspondingly, if these

will help in the quantitative assessment of therapy as well.

Acknowledgements

We thank Erzsébet Kaldenecker for her assistance in data

acquisition. The authors gratefully acknowledge the contribution

of J. C. Stam in providing the mathematical formulas outlined

in the Appendix. Grant support: OTKA Hungarian Research

FundT048338,NKFPHungarianNationalResearchFund5/007,

1B/020/04.

Appendix A. Calculation of the Synchronization likelihood

(Stam and van Dijk, 2002)

The Synchronization likelihood (SL) is a measure of general-

izedsynchronization(Rulkovetal.,1995)betweentwodynamical

systems X and Y. The first step in the computation of the SL is to

convert the time series xiand yirecorded from X and Yas a series

of state space vectors using the method of time-delay embedding

(Takens, 1981):

?

where L is the time lag, and m the embedding dimension. From a

time series ofN samples,N−(m×L) vectorscan bereconstructed.

State space vectors Yiare reconstructed in the same way.

SL is defined as the conditional likelihood that the distance

between Yiand Yjwill be smaller than a cut-off distance ry,

given that the distance between Xiand Xjis smaller than a cut-

off distance rx. In the case of maximal synchronization this

likelihood is 1; in the case of independent systems, it is a small,

but nonzero number, namely Pref. This number is the likelihood

that two randomly chosen vectors Y (or X) will be closer than

the cut-off distance r. In practice, the cut-off distance is chosen

such that the likelihood of random vectors being close is fixed at

Pref, which is chosen identically for X and for Y. To follow how

Prefis used to fix rxand rythe correlation integral is considered:

Xi¼ xi;xiþL;xiþ2?L;xiþ3?LN ; xiþ m?1

ð Þ?L

?

Cr¼

2

N N ? w

The correlation integral Cris the likelihood that two random-

ly chosen vectors X will be closer than r. The vertical bars

represent the Euclidean distance between the vectors. N is the

numberofvectors,wistheTheilercorrectionforautocorrelation

(Theiler, 1986) and θ is the Heaviside function: θ(X)=0 if

XN=0 and θ(X)=1 if Xb0. Rxis chosen so that Crx=Prefand ry

is chosen so that Cry=Pref. The SL between X and Y can be

formally defined as:

ðÞ

X

N

i¼1

X

N?w

j¼iþw

h r ?jXi? Xjj

??

:

SL ¼

2

N N ? w

ðÞPref

X

N

i¼1

X

N?w

j¼iþw

h rx?jXi? Xjj

??

h ry?jYi? Yjj

??

:

SL is a symmetric measure of the strength of synchronization

between X and Y (SLXY=SLYX).

Calculation of Omega-complexity

Omega-complexity characterizes the covariance-relations of

multichannel EEG data (Wackermann, 1999). After computing

the covariance-values between the electrode channels, a principal

component analysisis implemented to explore ifthere are mutual

factors (components) that can describe the variance of the vari-

ables. The eigenvalue of the components expresses the extent

to which the component is responsible for the variance of

the original variables. The Shannon-entropy of the normalized

eigenvalues equals the logarithm of Omega:

logX ¼ ?

where λiis the normalized eigenvalue of the i-th component.

The formula reveals that if one component can describe all

the time series, it means that a single source (“mechanism”)

would be responsible for the dynamics of the EEG, in which

X

kilogki

ðÞ;

79B. Czigler et al. / International Journal of Psychophysiology 68 (2008) 75–80

Page 6

case logΩ=0 and thus Omega equals 1. If independent signals

are recorded in the channels (e.g. white-noise), then Omega

equals the number of channels. A low Omega value implies a

more simple, and a high Omega a more complex signal.

References

Abasolo, D., Hornero, R., Gomez, C., Garcia, M., Lopez, M., 2006. Analysis of

EEG background activity in Alzheimer's disease patients with Lempel–Ziv

complexity andcentraltendency measure.Medical Engineering & Physics28,

315–322.

Adler, G., Brassen, S., Jajcevic, A., 2003. EEG coherence in Alzheimer's

dementia. Journal of Neural Transmission 110, 1051–1058.

Alloul, K., Sauriol, L., Kennedy, W., Laurier, C., Tessier, G., Novosel, S.,

Contandriopoulos, A., 1998. Alzheimer's disease: a review of the disease, its

epidemiologyandeconomicimpact.ArchivesofGerontologyandGeriatrics27,

189–221.

Bennys, K., Rondouin, G., Vergnes, C., Touchon, J., 2001. Diagnostic value of

quantitative EEG in Alzheimer's disease. Neurophysiologie Clinique-Clinical

Neurophysiology 31, 153–160.

Blennow, K., de Leon, M.J., Zetterberg, H., 2006. Alzheimer's disease. Lancet

368, 387–403.

Braak, H., Braak, E., 1995. Staging of Alzheimer's disease-related neurofi-

brillary changes. Neurobiology of Aging 16, 271–278.

Braak, H., Del Tredici, K., 2004. Alzheimer's disease: intraneuronal alterations

precede insoluble amyloid-beta formation. Neurobiology of Aging 25,

713–718.

Brenner, R., 1999. EEG and Dementia, In: Niedermeyer, E., Lopes da Silva, F.

(Eds.), Electroencephalography: Basic Principles, Clinical Applications and

Related Fields, 4th ed. Williams & Wilkins, Baltimore, pp. 349–359.

Buzsaki, G., Draguhn, A., 2004. Neuronal oscillations in cortical networks.

Science 304, 1926–1929.

Coben, L.A., Danziger, W., Storandt, M., 1985. A longitudinal EEG study of

mildseniledementiaofAlzheimerType—changesat1yearandat2.5years.

Electroencephalography and Clinical Neurophysiology 61, 101–112.

Dringenberg,H.C.,2000.Alzheimer'sdisease:morethana‘cholinergicdisorder’—

evidence that cholinergic–monoaminergic interactions contribute to EEG

slowing and dementia. Behavioural Brain Research 115, 235–249.

Folstein,M.F.,Robins,L.N.,Helzer,J.E.,1983.Themini-mentalstateexamination.

Archives of General Psychiatry 40, 812.

Hachinski, V.C., Lassen, N.A., Marshall, J., 1974. Multi-infarct dementia. A

cause of mental deterioration in the elderly. Lancet 2, 207–210.

Hamilton,M.,1960.Aratingscalefordepression.JournalNeurology,Neurosurgery

and Psychiatry 23, 56–62.

Hogan, M.J., Swanwick, G.R.J., Kaiser, J., Rowan, M., Lawlor, B., 2003.

Memory-related EEG power and coherence reductions in mild Alzheimer's

disease. International Journal of Psychophysiology 49, 147–163.

Jeong, J., Gore, J.C., Peterson, B.S., 2001. Mutual information analysis of the

EEG in patients with Alzheimer's disease. Clinical Neurophysiology 112,

827–835.

Jeong, J.S., 2004. EEG dynamics in patients with Alzheimer's disease. Clinical

Neurophysiology 115, 1490–1505.

Klimesch, W., 1999. EEG alpha and theta oscillations reflect cognitive and

memory performance: a review and analysis. Brain ResearchBrain Research

Reviews 29, 169–195.

Klimesch, W., Sauseng, P., Hanslmayr, S., 2007. EEG alpha oscillations: the

inhibition-timing hypothesis. Brain Research Reviews 53, 63–88.

Koenig, T., Prichep, L., Dierks, T., Hubl, D., Wahlund, L.O., John, E.R., Jelic,

V., 2005. Decreased EEG synchronization in Alzheimer's disease and mild

cognitive impairment. Neurobiology of Aging 26, 165–171.

Locatelli, T., Cursi, M., Liberati, D., Franceschi, M., Comi, G., 1998. EEG

coherence in Alzheimer's disease. Electroencephalography and Clinical

Neurophysiology 106, 229–237.

McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., Stadlan, E.M.,

1984. Clinical diagnosis of Alzheimer's disease: report of the NINCDS–

ADRDAWorkGroup under theauspices of Department of HealthandHuman

Services Task Force on Alzheimer's Disease. Neurology 34, 939–944.

Pijnenburg, Y.A.L., Made, Y.V., van Walsum, A.M.V., Knol, D.L., Scheltens, P.,

Stam,C.J.,2004.EEGSynchronizationlikelihoodinmildcognitiveimpairment

and Alzheimer's disease during a working memory task. Clinical Neurophy-

siology 115, 1332–1339.

Robinson, D.J., Merskey, H.,Blume, W.T., Fry, R., Williamson,P.C.,Hachinski,

V.C.,1994.ElectroencephalographyasanaidintheexclusionofAlzheimer's

disease. Archives of Neurology 51, 280–284.

Rulkov, N.F., Sushchik, M.M., Ysimring, L.S., Abarbanel, H.D.I., 1995.

Generalized synchronization of chaos in directionally coupled chaotic

systems. Physical Review E 51, 980–994.

Stam, C.J., 2005. Nonlinear dynamical analysis of EEG and MEG: review of an

emerging field. Clinical Neurophysiology 116, 2266–2301.

Stam, C.J., Jelles, B., Achtereekte, H.A.M., vanBirgelen, J.H., Slaets, J.P.J.,

1996. Diagnostic usefulness of linear and nonlinear quantitative EEG

analysisinAlzheimer'sdisease.ClinicalElectroencephalography27,69–77.

Stam,C.J.,vanDijk,B.W.,2002.Synchronizationlikelihood:anunbiasedmeasure

of generalized synchronization in multivariate data sets. Physica D-Nonlinear

Phenomena 163, 236–251.

Takens, F., 1981. Detecting strange attractors in turbulence. Lecture Notes in

Mathematics 898, 366–381.

Theiler, J., 1986. Spurious dimension from correlation algorithms applied to

limited time-series data. Physical Review A 34, 2427–2432.

van der Hiele, Vein, A.A., van der, W.A., van der, G.J., Westendorp, R.G.,

Bollen, E.L., van Buchem, M.A., van Dijk, J.G., Middelkoop, H.A., 2007.

EEG and MRI correlates of mild cognitive impairment and Alzheimer's

disease. Neurobiology of Aging 28, 1322–1329.

Vellas, B., Fitten, L., 2000. Research and Practice in Alzheimer's Disease.

Springer, New York.

Wackermann, J., 1999. Towards a quantitative characterisation of functional

states of the brain: from the non-linear methodology to the global linear

description. International Journal of Psychophysiology 34, 65–80.

Wong,P.,Li,T.,Price,D.,2006.NeurobiologyofAlzheimer'sdisease,In:Wong,

P.,Siegel,G.,Albers,R.,Brady,S.,Prichep,L.(Eds.),BasicNeurochemistry:

Molecular, Cellular and Medical Aspects, 7th ed. Elsevier, pp. 781–790.

Wurtman, R., Corkin, S., Growdon, J., Nitsch, R., 1996. The Neurobiology of

Alzheimer's Disease. The New York Academy of Sciences, New York.

80B. Czigler et al. / International Journal of Psychophysiology 68 (2008) 75–80