Induced oscillatory responses during the Sternberg's visual memory task in patients with Alzheimer's disease and mild cognitive impairment.

Abstract

In this study we used magnetoencephalography during a modified version of the Sternberg's memory recognition task performed by patients with early Alzheimer's disease (AD), mild cognitive impairment (MCI), and by age-matched healthy controls to identify differences in induced oscillatory responses. For analyses, we focused on the retention period of the working memory task. Multiple-source beamformer and Brain Voyager were used for localization of source-power changes across the cortex and for statistic group analyses, respectively. We found significant differences in oscillatory response during the task, specifically in beta and gamma frequency bands: patients with AD showed reduced beta event-related desynchronization (ERD) in the right central area compared to controls, and reduced gamma ERD in the left prefrontal and medial parietal cortex compared to patients with MCI. Our findings suggest that reduced oscillatory responses over certain brain regions in high frequency bands (i.e., beta, gamma), and especially in the beta band that was significantly different between AD patients and healthy subjects, may represent brain electromagnetic changes underlying visual-object working memory dysfunction in early AD, and a neurophysiological indicator of cognitive decline.

Full text

Page 1

Induced oscillatory responses during the Sternberg's visual memory task in patients
with Alzheimer's disease and mild cognitive impairment
Ryu Kurimoto1, Ryouhei Ishii⁎, Leonides Canuet1, Koji Ikezawa1, Masao Iwase1, Michiyo Azechi1,
Yasunori Aoki1, Shunichiro Ikeda1, Tetsuhiko Yoshida1, Hidetoshi Takahashi1, Takayuki Nakahachi1,
Hiroaki Kazui1, Masatoshi Takeda1
Department of Psychiatry, Osaka University Graduate School of Medicine, 2-2 D3, Yamadaoka, Suita, Osaka, 565-0871, Japan
a b s t r a c ta r t i c l e i n f o
\Article history:
Received 9 July 2011
Revised 18 October 2011
Accepted 18 October 2011
Available online 25 October 2011
Keywords:
Event-related desynchronization (ERD)
Magnetoencephalography (MEG)
Alzheimer's disease (AD)
Mild cognitive impairment (MCI)
Working memory
In this study we used magnetoencephalography during a modified version of the Sternberg's memory recog-
nition task performed by patients with early Alzheimer's disease (AD), mild cognitive impairment (MCI), and
by age-matched healthy controls to identify differences in induced oscillatory responses. For analyses, we fo-
cused on the retention period of the working memory task. Multiple-source beamformer and Brain Voyager
were used for localization of source-power changes across the cortex and for statistic group analyses, respec-
tively. We found significant differences in oscillatory response during the task, specifically in beta and gamma
frequency bands: patients with AD showed reduced beta event-related desynchronization (ERD) in the right
central area compared to controls, and reduced gamma ERD in the left prefrontal and medial parietal cortex
compared to patients with MCI. Our findings suggest that reduced oscillatory responses over certain brain
regions in high frequency bands (i.e., beta, gamma), and especially in the beta band that was significantly dif-
ferent between AD patients and healthy subjects, may represent brain electromagnetic changes underlying
visual-object working memory dysfunction in early AD, and a neurophysiological indicator of cognitive
decline.
© 2011 Elsevier Inc. All rights reserved.
Introduction
Alzheimer's disease (AD) is the most common type of dementia
and one of the main health problems in the elderly worldwide. Its
overall prevalence is more than 1% of the general population and
reaches 20% for those aged 80 or over (Ramaroson et al., 2003). AD
is characterized by neuronal degeneration in several brain regions
with widespread neuronal cell loss and the presence of neurofibril-
lary tangles and senile plaques. In the early stages of the disease,
the most commonly recognized symptom is memory loss, in particu-
lar difficulty in remembering recently learned facts, as well as prob-
lems with thinking and concentration. In addition, the concept of a
prodromal stage of AD, termed mild cognitive impairment (MCI),
has been proposed (Petersen, 2004). Recent studies have demonstrat-
ed that pharmacological treatment for early AD and MCI can slow the
progression of the disease (Feldman and Jacova, 2005). Therefore, an
early diagnosis and treatment appear to be a very important factor for
improving prognosis in patients with AD.
Magnetoencephalography (MEG) is a technique specifically
designed to measure neural activity noninvasively featuring high
time and spatial resolution. Unlike other neurophysiological tech-
niques (e.g., electroencephalography — EEG), MEG can directly detect
the electromagnetic activity of the brain without interferences of
skin, skull and cerebral fluid, which act as a low pass filter
(Hämäläinen, 1992). Thus, MEG is more suitable to explore brain os-
cillations, especially fast activity in beta and gamma bands, compared
to EEG. However, MEG has not extensively been used for diagnosis of
AD. Most MEG studies on AD and MCI have employed single dipole as
their core method of analysis (Fernández et al., 2002, 2006; Maestú
et al., 2001, 2006). This method has proven useful for studying focal
brain activity, particularly epileptic discharges. However, findings
from several neuroimaging and neurophysiological studies suggest
that a wide area of the brain is activated in higher information proces-
sing such as memorization (Cohen et al., 1997; Jokisch and Jensen,
2007), and that induced oscillatory activity may be the key to under-
standing functional communication in the brain, especially with re-
gard to memory and integrated functions (Başar et al., 2001; Ishii
et al., 2009; Pfurtscheller and Lopes da Silva, 1999). Thus, applying
MEG-dipole modeling, which identifies center of gravity rather than
the volume of activation might not be sufficient to visualize abnormal
activity in an extended network of sources underlying cognitive dys-
function in AD.
In the last decades, different approaches have been used to ana-
lyze MEG activity during cognitive task performance. For instance,
NeuroImage 59 (2012) 4132–4140
⁎ Corresponding author. Fax: +81 6 6879 3059.
E-mail address: ishii@psy.med.osaka-u.ac.jp (R. Ishii).
1Fax: +81 6 6879 3059.
1053-8119/$ – see front matter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2011.10.061
Contents lists available at SciVerse ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg

Page 2

minimum norm estimation (MNE) procedure has extensively been
used to estimate the cortical origin of the brain electromagnetic
response. MNE models have already been applied for source recon-
struction of MEG data in patients with dementia or in healthy elderly
with subjective memory complaints (Maestú et al., 2008, 2011).
Beamformers approach is a spatial filtering method which has also
become increasingly valuable for source reconstruction of MEG activ-
ity (Ishii et al., 1999; Robinson and Rose, 1992). Although beamfor-
mers analyses are unable to distinguish two sources if their time-
courses are 100% correlated, unlike MNE, they can easily handle
both superficial and deep sources, and a variety of statistical analyses
can be easily implemented (Huang et al., 2004). These approaches
have been most successful in identifying induced changes in cortical
oscillatory power that do not result in a strong average signal
(Hillebrand et al., 2005). Beamformers, in particular, have given us
an insight into the dynamics of oscillatory changes across the cortex
not explored previously with traditional analysis that rely on aver-
aged evoked responses (Ishii et al., 2009). By using MEG beamformer,
the topographic mapping of source-power changes across the brain is
obtained and locations with significant neuronal activation can be
detected (Chen et al., 2006). In particular, multiple source beamfor-
mer (MSBF), a modified version of the linearly constrained minimum
variance vector beamformer in the time frequency domain, has prov-
en to be of great value in the identification of oscillatory activity
source-power changes induced by sensory and cognitive tasks in
dementia (Kurimoto et al., 2008) and other neuropsychiatric disor-
ders such as psychosis (Canuet et al., 2010) and autism (Honaga
et al., 2010).
The assessment of EEG/MEG induced oscillatory response in dif-
ferent frequency bands in terms of power decrease or event-related
desynchronization (ERD) and power increase or event-related syn-
chronization (ERS) is a valuable way to reveal different aspects of
information processing in the normal and pathological brain
(Pfurtscheller and Lopes da Silva, 1999). Earlier ERD/ERS studies dur-
ing a memory task focused mainly on changes in theta (4–8 Hz),
alpha (8–15 Hz) and beta (15–30 Hz) frequency bands. Induced oscil-
latory responses in theta and alpha activity, for example, have been
reported associated with working memory (Bastiaansen et al., 2002;
Jensen and Tesche, 2002) and attention (Kahana et al., 2001), while
event-related beta oscillatory response, in particular beta ERD, exten-
sively investigated in relation to motor function, has been proposed
to underlie working memory processes, as well (Karrasch et al.,
2004; Pesonen et al., 2007). In addition, EEG and MEG research on
cognitive function has recently shifted to high frequency oscillations
(i.e., gamma frequency band). This fast cortical oscillatory activity is
thought to be associated with various cognitive processes, and there-
fore its alteration appears to be an important mechanism underlying
psychiatric and neurological disorders (Grabska-Barwińska and
Zygierewicz, 2006; Jokisch and Jensen, 2007; Uhlhaas and Singer,
2006).
Based on the fact that memory impairment is a cardinal clinical
feature of AD and MCI, and that task-induced oscillatory brain activity
in different frequency bands provide important clues to underlying
cognitive processes (Başar et al., 2001; Pfurtscheller and Lopes da
Silva, 1999), several studies have investigated abnormal event-
related responses during memory tasks in demented patients. An
MEG study by Babiloni et al. (2005) reported that patients with de-
mentia showed increased alpha ERD during the retention period of
a memory task. Meanwhile, Karrasch et al. (2006) study using EEG
found that patients with MCI showed increased ERD in the frequency
range of 10–20 Hz during the encoding period compared to controls,
whereas patients with AD showed decreased ERD in the 7–17 Hz fre-
quencies during the retrieval period compared to controls. This activ-
ity was observed particularly in anterior and left temporal electrodes
(Karrasch et al., 2006). In addition, another EEG study reported de-
creased 15–25 Hz ERS in parietal electrodes during a 2-back task in
patients
(Missonnier et al., 2007). However, there are only few functional im-
aging studies evaluating ERD/ERS during a memory task in patients
with AD and MCI in the frequency range from theta to gamma.
In the present study, we used MEG beamformer during a visual
working memory task in patients with early AD and compared the re-
sults with those of patients with MCI and normal controls using Brain
Voyager to identify abnormal spatial patterns of oscillatory activity in
the wide frequency range.
withprogressiveMCIandADrelativetocontrols
Methods
Subjects
Thirteen patients with early AD, thirteen with amnestic MCI, and
fourteen normal elderly controls were enrolled in this study. All pa-
tients were recruited from the outpatient clinic of Psychiatry at
Osaka University Hospital. The study was carried out in accordance
with the Declaration of Helsinki, and approved by the Hospital Ethics
Committee. Written informed consent was obtained from all partici-
pants. The diagnosis of probable AD was established according to
the National Institute of Neurological and Communicative Disorders
and Stroke/the Alzheimer's Disease and Related Disorders Association
(NINCDS-ADRDA) criteria (McKhann et al., 1984), while amnestic
MCI was diagnosed according to the criteria defined by Petersen
(Petersen, 2004). The elderly controls were healthy volunteers who
had no cognitive disturbance and no history of neurological or psychi-
atric disorders. Patients and controls were not taking any medication
that might affect the central nervous system at the time of the study,
and underwent brain MRI screening to exclude any organic lesion. To
evaluate the degree of cognitive function, the Mini-Mental State Ex-
amination (MMSE) (Folstein et al., 1975) was performed on all pa-
tients and controls. In addition, the Alzheimer's Disease Assessment
Scale-cognitive subscale (ADAS-Cog) (Homma et al., 1992) and the
Clinical Dementia Rating (CDR) (Morris, 1993) were performed on
all patients. Their demographic and neuropsychological profiles are
shown in Table 1.
MEG data acquisition
Neuromagnetic data were recorded at 250 Hz with a bandwidth of
0–80 Hz using a CTF 64-channel MEG system (CTF Systems, Inc., Can-
ada) composed of a whole-head array of 64 radial 1st order gradiom-
eter/SQUID channels housed in a magnetically shielded room. The
participants were seated comfortably with the head positioned in
the helmet-shaped Dewar. The localization of the subject's head rela-
tive to the sensor array was measured with three coils affixed to the
nasion and preauricular points.
Table 1
Demographic, clinical and behavioral data.
AD MCIcontrolsp
Number
Sex (male/female)
Age
MMSE
ADAS-Cog
CDR
Accuracy rate of
Sternberg's task (%)
13
4/9
75.6±5.0
22.1±2.6
14.3±3.3
0.8±0.2
80.5±13.8
13
5/8
73.9±5.0
26.8±2.0
8.6±3.3
0.4±0.2
88.4±7.7
14
6/8
71.2±6.8
28.6±1.5
–
–
90.6±7.7
0.81
0.25
b0.001a,b
b0.001b
b0.001b
0.05a,b
Data are means±SD unless otherwise noted.
AD, Alzheimer's disease; MCI, mild cognitive impairment; MMSE, Mini-Mental State
Examination; ADAS-Cog, Alzheimer's Disease Assessment Scale-cognitive subscale;
CDR, Clinical Dementia Rating.
aComparison between AD and controls.
bComparison between AD and MCI.
4133
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 3

Working memory task
A modified version of the Sternberg's probe task was performed
during the MEG recording (Sternberg, 1966) (Fig. 1). During the
task, sets of five digits, or memory sets, were randomly displayed
for 2 s and the subjects were asked to memorize them (memory
encoding). After a 5-second memory retention period following the
memory set, a series of three single-digit probes (1-s duration with
a 2-s interval) was displayed and the subjects were instructed to
push a button held in their right hand if the digit was included in
the previous set, otherwise they had to push a button in their left
hand (memory recognition). The length of a single trial was 27 s
and repeated 20 times, resulting in a total duration of the task of
approximately 9 min per subject. Before the experiment, all subjects
were givencompletetaskinstructions;practice trialswere
performed to ensure familiarity with the procedure. The subject's
recall percentage was determined and deemed task performance.
For analyses, we focused on the retention period of the working
memory task (see MEG time-frequency analyses below).
MEG time-frequency analyses
Artifact rejection was performed off-line. MEG channels and trials
with signal variations larger than 3pT were considered as including
artifacts and were excluded from further analysis. For source imaging
of MEG data, we used the MSBF implemented in Brain Electrical
Source Analysis (BESA) software (www.besa.de) that represents a
modified version of the linearly constrained minimum variance vec-
tor beamformer in the time-frequency domain (Gross et al., 2001).
As an adaptive beamformer, the MSBF applies a spatial filter specific
Fig. 1. Schematic representation of an example trial of the Sternberg's visual memory task. Each trial included the presentation of a string of five digits (memory set) for 2 s (3 s after
the “START” cue); a 5-second retention interval, and the sequential presentation of three single probe digits of 1-second duration. The control and active time-windows are indi-
cated by double-head arrows. A cue “+” symbol was displayed during the retention period, and between trials and visual stimuli.
Fig. 2. Averaged source-power changes in the theta frequency band (4–8 Hz) across all groups at a threshold of 10%. The color bars represent the percentages of decreased power or
event-related desynchronization (blue/green, ERD) and increased power or event-related synchronization (red/yellow, ERS). L, left; R, right; A, anterior; P, posterior; TRA, trans-
verse; SAG, sagittal; COR, coronal; MCI, mild cognitive impairment; AD, Alzheimer's disease.
4134
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 4

for each brain voxel that is fully sensitive to activity from the target
voxel, while being as insensitive as possible to activity from other
brain regions, thus suppressing interference from unwanted signals.
The BESA beamformer applied complex demodulation to transform
time-domain MEG data into time-frequency data. This provided in-
formation on the envelope amplitude and the phase of a specified fre-
quency band as a function of time (Hoechstetter et al., 2004). The
complex demodulation consisted of a multiplication of the time-
domain signal by a complex periodic potential function with a fre-
quency equal to the frequency analyzed and an additional low-pass
filter. In the resulting complex signal, its magnitude corresponded
to half the envelope amplitude and its phase to the compound
phase of the filtered frequency band. To obtain power values, the
time-series MEG data were squared and averaged across all 20 trials.
Time frequency representations of changes in power normalized to
baseline for each MEG sensor were obtained from each subject.
Task-inducedoscillatoryresponses
(4–8 Hz), alpha (8–15 Hz), beta (15–30 Hz) and gamma (30–80 Hz)
bands. For each trial, a 2-s interval before the memory set (time
window: 1–3 s after the trigger) was deemed control state, whereas
an interval of equal duration starting 1-s after the memory set in
the memory retention period (time window: 6–8 s after the trigger)
was deemed active state (Fig. 1). To image induced oscillatory
activity, BESA computed the complex cross spectral density matrices
(the time-frequency equivalent of the data covariance matrix) for
the active and control intervals from single-trial data for each
were measured intheta
frequency band of interest. Color-coded maps were obtained display-
ing q-values as a measure of the magnitude change in the active inter-
val relative to the control interval in percentage.
Statistical group analyses
For statistical group analyses, the 3D images of induced oscillatory
activity built with BESA software were exported to the Brain Voyager
QX software package (Brain Innovation, Maastricht, The Nether-
lands), and superimposed onto a Talairach-transformed Montreal
Neurological Institute (MNI) T1-weighted brain MRI template in
Brain Voyager QX (Goebel et al., 2006). The anatomical T1 coordi-
nates in the statistic maps were transformed into Talairach coordi-
nates to identify brain regions with significant between-group
differences in source-power changes. Brain activation patterns, as in-
dicated by ERD/ERS values in a given frequency band, were compared
between two groups using t-test statistics (two-tailed, unpaired) in
BrainVoyager QX. To minimize the risk of false positive findings, all
activation foci were set to a minimum cluster size of 20 voxels, and
statistical results with pb0.001 (uncorrected) were considered signif-
icant. Details of these neuroimaging procedures can be found in our
recent publications (Canuet et al., 2010, Honaga et al., 2010;
Kurimoto et al., 2008). The Chi-square test was performed for inde-
pendence of group and gender. Demographic and clinical variables
were analyzed using the Mann Whitney U test or the Kruskal–Wallis
Fig. 3. Averaged source-power changes in the alpha band (8–15 Hz) across all groups at a threshold of 10%. The color bars represent the percentages of decreased power or event-
related desynchronization (blue/green, ERD) and increased power or event-related synchronization (red/yellow, ERS). L, left; R, right; A, anterior; P, posterior; TRA, transverse; SAG,
sagittal; COR, coronal; MCI, mild cognitive impairment; AD, Alzheimer's disease.
4135
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 5

test, with the significance level set at pb0.05. These statistical ana-
lyses were carried out using SPSS software (SPSS, Inc., Chicago, USA).
Results
Demographic, clinical and behavioral results
The demographic data are shown in Table. 1. No significant differ-
ences were indicated in either age (d.f. = 38, p=0.25) or sex (d.f. =
2, T=0.43, p=0.81) across the three groups. The neuropsychological
assessment revealed significant differences in MMSE scores between
AD and MCI patients (Kruskal–Wallis test, pb0.01), as well as be-
tween AD patients and controls (Kruskal–Wallis test, pb0.01). For
the ADAS-Cog and CDR scores a significant difference was found be-
tween AD and MCI patients (Mann–Whitney test, pb0.001). All sub-
jects achieved an accuracy rate of 70% or higher on the Sternberg's
task during MEG recording. There was significant difference in the
correct-response percentage only between AD patients and normal
controls, with the patients showing a lower performance (Kruskal–
Wallis test, pb0.05).
Within-group power change analysis
The averaged percentages of power changes in different frequency
bands for each group, superimposed on a standard brain image are
shown in Figs. 2–5. Pronounced theta ERD exceeding 10% was ob-
served in the left frontocentral and inferior temporal region in both
AD patients and normal controls. Maximal activity (peak ERD) was
found in the frontal cortex with values of 11.2% in the patients and
10.4% in the controls. In MCI patients, pronounced theta ERD was ob-
served in a wide area including the frontocentral cortex bilaterally
and the left posterior temporal cortex, with a peak ERD value of
12.8% in the frontal area, as well as in the contralateral frontal cortex
(peak ERD 11.3%) (Fig. 2).
Alpha power showed marked reduction over a wide cortical area
involving the left frontocentral cortex in all three groups, and the ad-
jacent superior temporal region in MCI patients and controls. Maxi-
mal ERD was observed in the left frontal cortex in all groups with
values of 15.3%, 15.5% and 12.5% in controls, and patients with MCI
and AD, respectively. In MCI patients there was also alpha ERD greater
than 10% in the right frontal cortex (peak ERD 13.6%) (Fig. 3).
In the beta frequency band, pronounced ERD was observed only in
MCI patients and controls. This activity was widely distributed over
the frontocentral cortex bilaterally, dominant over the left hemi-
sphere. Maximal beta ERD was found in the left frontal cortex in
both groups, with peak values of 16.4% in controls and 15.6% in MCI
patients. The maximum value of beta ERD in AD patients was 9%
and is was also found over the left frontal region (Fig. 4).
Gamma frequency band showed no power change greater than
10%. At a threshold of 5% power changes, gamma ERD was observed
Fig. 4. Averaged source-power changes in the beta band (15–30 Hz) across all groups at a threshold of 10%. The color bars represent the percentages of decreased power or event-
related desynchronization (blue/green, ERD) and increased power or event-related synchronization (red/yellow, ERS). L, left; R, right; A, anterior; P, posterior; TRA, transverse; SAG,
sagittal; COR, coronal; MCI, mild cognitive impairment; AD, Alzheimer's disease.
4136
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 6

in all groups. This activity appeared mainly in the frontocentral cortex
bilaterally in normal controls, with a peak value of 8.1% in the right
frontal region, and in MCI patients, in the left frontocentral and medi-
al parietal cortex, with a peak value of 8.5% over the prefrontal area.
Patients with AD showed gamma reduction greater than 5% in a
small area over the right prefrontal cortex, with a peak value of 6.2%
(Fig. 5). Pronounced ERS was not observed in any of the frequency
bands.
Between-group power change analysis
The between-group comparisons of source-power changes in
brain oscillatory activity during the retention period of the Stern-
berg's working memory task indicated significant difference in beta
and gamma frequency bands only. Patients with AD had reduced
magnitude of beta ERD in the right central area relative to controls,
with tmaximain the precentral cortex (Fig. 6). Patients with AD also
Fig. 5. Averaged source-power changes in the gamma band (30–80 Hz) across all groups at a threshold of 6%. The color bars represent the percentages of decreased power or event-
related desynchronization (blue/green, ERD) and increased power or event-related synchronization (red/yellow, ERS). L, left; R, right; A, anterior; P, posterior; TRA, transverse; SAG,
sagittal; COR, coronal; MCI, mild cognitive impairment; AD, Alzheimer's disease.
Fig. 6. Statistical maps showing cortical regions with significant differences in beta (15–30 Hz) power changes during working memory retention between patients with
Alzheimer's disease versus healthy controls projected onto a Talairach-transformed T1-weighted anatomical MRI (threshold: t=3.73, Pb0.001). A significant decrease in beta
ERD was shown in the right central area in patients with Alzheimer's disease relative to healthy controls. L, left; R, right; A, anterior; P, posterior; TRA, transverse; SAG, sagittal;
COR, coronal.
4137
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 7

showed reduced magnitude of gamma ERD relative to MCI patients,
specifically in the left dorsolateral prefrontal cortex (DLPFC) over
the superior frontal gyrus, and in the left medial parietal cortex (i.e.,
precuneus), with tmaximain the DLPFC source (Fig. 7). The findings
in peak beta and gamma ERD values at tmaximaare summarized in
Table 2.
Discussion
In the present study, we compared induced oscillatory activity
during the retention period of a modified version of the Sternberg's
memory recognition task in patients with early AD and MCI, and in
normal controls in an attempt to detect early brain electromagnetic
changes underlying memory impairment in AD. We found pro-
nounced power changes (i.e., decrease in power or ERD) exceeding
10% in theta, alpha and beta frequency bands in patients and controls,
with a similar topographic pattern. This activity was distributed
mainly over the left frontocentral area, and the left temporal cortex
to a lesser extent, with AD patients showing smaller cortical areas
of activation than the other groups. We also noted ERD source over
the contralateral frontal cortex in these three frequency bands in pa-
tients with MCI, and it was also seen in normal controls but only for
the beta band. Gamma activity exhibited a small magnitude of
power changes across the cortex in all three groups. However, the
statistical group comparison using Brain Voyager indicated significant
differences (pb0.001) in induced oscillatory response during the task
in beta and gamma frequency bands: patients with AD showed re-
duced beta ERD in the right central area compared to controls, and re-
duced gamma ERD in the left prefrontal and left precuneus compared
to patients with MCI.
Beta ERD/ERS responses have extensively been studied in relation
to motor processing (Pfurtscheller and Klimesch, 1991). Recent stud-
ies, however, indicate that power changes in the beta frequency band,
in particular beta ERD also appear to play an important role in cogni-
tive functioning, including working memory processing (Pesonen
et al., 2006), with activation in the beta band correlating positively
with memory load (Pesonen et al., 2007). This is consistent with our
finding of significant decrease in beta ERD in AD patients relative to
normal controls during performance of a working memory task
(Table 2, Fig. 6). In contrast to our result of beta ERD decrease in AD
during memory retention, Missonnier et al. reported decreased beta
ERS in AD and progressive MCI during a 2-back task (Missonnier
et al., 2007). The discrepancy between these two studies can be
explained, at least in part, by the fact that the n-back task is thought
to require memory encoding, retention and retrieval simultaneously,
while we explored abnormalities related exclusively to the retention
period of the Sternberg's task. In the present study, the abnormal
beta oscillatory activity found in AD patients was distributed over
the right frontocentral region, specifically in the precentral gyrus as
well as in the lateral prefrontal cortex over the middle frontal gyrus,
which is known to subserve working memory operations (Baddeley,
2003). Results from a PET study demonstrating that regional metabol-
ic deficits in the prefrontal association cortex, as indicated by FDG up-
take decline, was linked to the severity of dementia and discriminated
patients with early AD from healthy subjects (Herholz et al., 2002)
provides support to the beta power source-localization finding of
our study.
Power changes greater than 10% were not observed in the gamma
frequency band for any of the studied groups. However, the statistical
analyses revealed significantly decreased gamma ERD during working
memory-retention in AD patients relative to MCI patients (Table 2).
This activity located to the left prefrontal cortex over the middle fron-
tal lobe and to the left precuneus in the medial aspect of parietal lobe.
There was no significant difference in gamma oscillatory response
when comparing patients with healthy controls. Induced gamma os-
cillations are thought to reflect feature-binding processes, and to
generate a neural representation of the specific stimuli used (e.g.,
cognitive stimulus) (Herrmann and Demiralp, 2005). Thus, our find-
ings in the gamma band might reflect stimulus-related abnormalities
among patients with early AD. This assumption is consistent with
Fig. 7. Statistical maps showing cortical regions with significant differences in gamma (30–80 Hz) power changes during working memory retention between patients with Alzhei-
mer's disease versus mild cognitive impairment projected onto a Talairach-transformed T1-weighted anatomical MRI (threshold: t=3.75, Pb0.001). A significant decrease in
gamma ERD was shown in the left prefrontal and medial parietal cortex in patients with Alzheimer's disease relative to patients with mild cognitive impairment. L, left; R, right;
A, anterior; P, posterior; TRA, transverse; SAG, sagittal; COR, coronal.
Table 2
Cortical regions showing significant between-group differences in oscillatory activity power changes.
ERD LocationTalairach
(x, y, z)
BAERD at Tmaxima (%)T-value
AD MCI controls
Beta ERD
Gamma ERD
Gamma ERD
Right MFG
Left SFG
Left PreC
46, 5, 52
−24, 9, 58
−11 −65, 44
6
6
7
−1.45±4.39
−1.62±2.97
−0.66±2.73
−4.36±3.61
−7.11±3.99
−6.18±4.00
−8.68±4.71
−3.75±2.94
−2.32±3.24
3.831a
3.863b
4.000b
ERD values are means±SD.
ERD, event-related desynchronization; BA, Brodmann's area; AD, Alzheimer's disease; MCI, mild cognitive impairment; MFG, middle frontal gyrus; SFG, superior frontal gyrus; PreC,
precuneus.
aComparison between AD and controls.
bComparison between AD and MCI.
4138
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 8

reports of decreased gamma oscillations in AD, in particular over the
prefrontal region which is linked to working memory processing, and
with findings from previous studies indicating that gamma activity is
particularly affected by aging, probably due to age-dependent loss of
dopamine D2 receptors in cortical circuits (Li et al., 2001). In addition
to the left prefrontal cortex, the ipsilateral precuneus also showed
gamma abnormalities in patients with AD that differentiated them
from those with MCI. This supports evidence from functional imaging
studies indicating that hypoperfusion or hypometabolism in the pre-
cuneus might be an early functional marker of AD (Herholz et al.,
2002; Kogure et al., 2002). Furthermore from a different angle, it
could be said that patients with MCI tended to show higher gamma
ERD compared to AD and controls (Table 2). This tendency might re-
flect a compensatory mechanism that allows patients with MCI to
achieve similar behavioral results to those of controls during a mem-
ory task, as suggested by findings from previous MEG studies (Bajo
et al., 2010; Maestú et al., 2008).
Several MEG studies reported significant differences in oscillatory
activity between MCI and controls during a memory task (Maestú
et al., 2006, 2008, 2011). In our study, however, the comparison be-
tween these two groups showed no statistical significance. The fact
that we could not divide MCI patients into subgroups may partly ex-
plain the difference in the findings. Interestingly, results from recent
studies looking at MCI progression to AD have demonstrated that
some patients with MCI do not progress to AD (Petersen, 2004), and
in an MEG study MCI patients were divided into converters who pro-
gressed to AD, and non-converters who did not progress to AD
(Maestú et al., 2006). Follow up studies and subgroup analyses will
help clarify the significant differences in source-power changes be-
tween MCI and controls.
A number of studies suggest a role for other frequency bands, in
particular theta and alpha bands, in working memory processing
(Pesonen et al., 2006, 2007; Stam, 2000), and reported that abnormal-
ities in these bands are commonly involved in working memory def-
icits in several neuropsychiatric diseases such as schizophrenia and
dementia (Babiloni et al., 2005; Ince et al., 2009; Montez et al.,
2009). While we applied a visual-object working memory task using
digits, most previous studies, however, used different working mem-
ory tasks (e.g., auditory memory search, visual–spatial or verbal
memory tasks). This raises the question as to whether the discrepan-
cy in the findings concerning the frequency bands across studies
might be due to difference in working-memory paradigm and
disease-related factors. Overall, our findings suggest that reduced os-
cillatory responses in high frequency bands (i.e., beta, gamma), and
especially in the beta band that was significantly different between
AD patients and age-matched healthy subjects, may reflect neural
activity underlying visual-object working memory dysfunction in
early AD.
Our findings should be interpreted in the context of potential lim-
itations, including a small sample size and the brain atrophy of the
patients that was not considered. Nevertheless, all patients groups
and healthy controls were carefully matched by age and sex, and
those with structural brain lesions were not included. In addition,
since all AD patients in this study were in the early stage of the dis-
ease, and consequently did not show severe brain volume reduction,
our results are unlikely to be affected by the brain atrophy factor. Fur-
ther larger studies with a longitudinal design may help clarify the
specific role of beta and gamma power changes in terms of ERD/ERS
as a neurophysiological indicator of cognitive decline in AD. It should
also be noted that we did not carry out separate analysis for the MEG
data of correct and incorrect answers but averaged the data together.
This was done because only 24 probes were presented in our version
of Sternberg's memory task, and even patients with AD performed
quite well on the task, having an average accuracy rate of 80%. Larger
sample size would provide enough statistical power to investigate the
differences of MEG activities between correct and incorrect answers.
Finally, we should consider the relatively small number of channels
of the MEG system used, as more than 100 channels is standard for
MEG systems nowadays. However, striking findings of recent studies
using MEG equipped with 64-channels indicating an accurate locali-
zation of auditory evoked fields (Johnson et al., 2010) as well as of os-
cillatory activity associated with language dominance (Hirata et al.,
2010) and brain tumors (Oshino et al., 2007) speak in favor of the
neuroimaging quality of this MEG system.
Funding sources
This study was supported in part by the Grant-in-Aid for Scientific
Research (No. 21591516) from the Japan Society for the Promotion of
Science (JSPS).
Conflicts of interest
None.
References
Babiloni, C., Cassetta, E., Chiovenda, P., Del Percio, C., Ercolani, M., Moretti, D.V., Moffa,
F., Pasqualetti, P., Pizzella, V., Romani, G.L., Tecchio, F., Zappasodi, F., Rossini, P.M.,
2005. Alpha rhythms in mild dements during visual delayed choice reaction time
tasks: a MEG study. Brain Res. Bull. 65 (6), 457–470.
Baddeley, A., 2003. Working memory: looking back and looking forward (review). Nat.
Rev. Neurosci. 4 (10), 829–839.
Bajo, R., Maestú, F., Nevado, A., Sancho, M., Gutiérrez, R., Campo, P., Castellanos, N.P.,
Gil, P., Moratti, S., Pereda, E., Del-Pozo, F., 2010. Functional connectivity in mild
cognitive impairment during a memory task: implications for the disconnection
hypothesis. J. Alzheimers Dis. 22 (1), 183–193.
Başar, E., Başar-Eroglu, C., Karakaş, S., Schürmann, M., 2001. Gamma, alpha, delta, and
theta oscillations govern cognitive processes. Int. J. Psychophysiol. 39 (2–3),
241–248.
Bastiaansen, M.C., Posthuma, D., Groot, P.F., de Geus, E.J., 2002. Event-related alpha and
theta responses in a visuo-spatial working memory task. Clin. Neurophysiol. 113
(12), 1882–1893.
Canuet, L., Ishii, R., Iwase, M., Ikezawa, K., Kurimoto, R., Takahashi, H., Currais, A.,
Azechi, M., Nakahachi, T., Hashimoto, R., Takeda, M., 2010. Working memory ab-
normalities in chronic interictal epileptic psychosis and schizophrenia revealed
by magnetoencephalography. Epilepsy Behav. 17 (1), 109–119.
Chen, Y.S., Cheng, C.Y., Hsieh, J.C., Chen, L.F., 2006. Maximum contrast beamformer for
electromagnetic mapping of brain activity. IEEE Trans. Biomed. Eng. 53 (9),
1765–1774.
Cohen, J.D., Perlstein, W.M., Braver, T.S., Nystrom, L.E., Noll, D.C., Jonides, J., Smith, E.E.,
1997. Temporal dynamics of brain activation during a working memory task.
Nature 386 (6625), 604–608.
Feldman, H.H., Jacova, C., 2005. Mild cognitive impairment. Am. J. Geriatr. Psychiatry 13
(8), 645–655.
Fernández, A., Maestú, F., Amo, C., Gil, P., Fehr, T., Wienbruch, C., Rockstroh, B., Elbert,
T., Ortiz, T., 2002. Focal temporoparietal slow activity in Alzheimer's disease
revealed by magnetoencephalography. Biol. Psychiatry 52 (7), 764–770.
Fernández, A., Turrero, A., Zuluaga, P., Gil, P., Maestú, F., Campo, P., Ortiz, T., 2006.
Magnetoencephalographic parietal delta dipole density in mild cognitive impair-
ment: preliminary results of a method to estimate the risk of developing
Alzheimer disease. Arch. Neurol. 63 (3), 427–430.
Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. “Mini-mental state”. A practical meth-
od for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12
(3), 189–198.
Goebel, R., Esposito, F., Formisano, E., 2006. Analysis of functional image analysis con-
test (FIAC) data with brainvoyager QX: From single-subject to cortically aligned
group general linear model analysis and self-organizing group independent com-
ponent analysis. Hum. Brain Mapp. 27 (5), 392–401.
Grabska-Barwińska, A., Zygierewicz, J., 2006. A model of event-related EEG synchroni-
zation changes in beta and gamma frequency bands. J. Theor. Biol. 238 (4),
901–913.
Gross, J., Kujala, J., Hämäläinen, M.S., Timmermann, L., Schnitzler, A., Salmelin, R., 2001.
Dynamic imaging of coherent sources: studying neural interactions in the human
brain. Proc. Natl. Acad. Sci. 98 (2), 694–699.
Hämäläinen, M.S., 1992. Magnetoencephalography: a tool for functional brain imaging.
Brain Topogr. 5 (2), 95–102.
Herholz, K., Salmon, E., Perani, D., Baron, J.C., Holthoff, V., Frölich, L., Schönknecht, P.,
Ito, K., Mielke, R., Kalbe, E., Zündorf, G., Delbeuck, X., Pelati, O., Anchisi, D., Fazio,
F., Kerrouche, N., Desgranges, B., Eustache, F., Beuthien-Baumann, B., Menzel, C.,
Schröder, J., Kato, T., Arahata, Y., Henze, M., Heiss, W.D., 2002. Discrimination be-
tween Alzheimer dementia and controls by automated analysis of multicenter
FDG PET. Neuroimage 17 (1), 302–316.
Herrmann, C.S., Demiralp, T., 2005. Human EEG gamma oscillations in neuropsychiatric
disorders. Clin. Neurophysiol. 116 (12), 2719–2733.
4139
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140

Page 9

Hillebrand, A., Singh, K.D., Holliday, I.E., Furlong, P.L., Barnes, G.R., 2005. A new ap-
proach to neuroimaging with magnetoencephalography. Hum. Brain Mapp. 25
(2), 199–211.
Hirata, M., Goto, T., Barnes, G., Umekawa, K., Yanagisawa, T., Kato, A., Oshino, S.,
Kishima, H., Hashimoto, N., Saitoh, Y., Tani, N., Yorifuji, S., Yoshimine, T., 2010.
Language dominance and mapping based on neuromagnetic oscillatory changes:
comparison with invasive procedures. J. Neurosurg. 112 (3), 528–538.
Hoechstetter, K., Bornfleth, H., Weckesser, D., Ille, N., Berg, P., Scherg, M., 2004. BESA
source coherence : a new method to study cortical oscillatory coupling. Brain
Topogr. 16 (4), 233–238.
Homma, A., Fukuzawa, K., Tsukada, Y., Ishii, T., Hasegawa, K., Mohs, R.C., 1992.
Development of a Japanese version of Alzheimer's disease Assessment Scale
(ADAS). Jpn. J. Geriatr. Psychiatry 3, 647–655 (in Japanese).
Honaga, E., Ishii, R., Kurimoto, R., Canuet, L., Ikezawa, K., Takahashi, H., Nakahachi, H.,
Iwase, M., Mizuta, I., Yoshimine, T., Takeda, M., 2010. Post-movement beta rebound
abnormality as indicator of mirror neuron system dysfunction in autistic spectrum
disorder: an MEG study. Neurosci. Lett. 478 (3), 141–145.
Huang, M.X., Shih, J.J., Lee, R.R., Harrington, D.L., Thoma, R.J., Weisend, M.P., Hanlon, F.,
Paulson, K.M., Li, T., Martin, K., Millers, G.A., Canive, J.M., 2004. Commonalities and
differences among vectorized beamformers in electromagnetic source imaging.
Brain Topogr. 16 (3), 139–158.
Ince, N.F., Pellizzer, G., Tewfik, A.H., Nelson, K., Leuthold, A., McClannahan, K., Stephane,
M., 2009. Classification of schizophrenia with spectro-temporo-spatial MEG pat-
terns in working memory. Clin. Neurophysiol. 120 (6), 1123–1134.
Ishii, R., Shinosaki, K., Ukai, S., Inoue, T., Ishihara, T., Yoshimine, T., Hirabuki, N., Asada,
H., Kihara, T., Robinson, S.E., Takeda, M., 1999. Medical prefrontal cortex generates
frontal midline theta rhythm. Neuroreport 10 (4), 675–679.
Ishii, R., Canuet, L., Herdman, A., Gunji, A., Iwase, M., Takahashi, H., Nakahachi, T., Hirata,
M.,Robinson,S.E.,Pantev,C.,Takeda,M.,2009.Corticaloscillatorypowerchangesdur-
ing auditory oddball task revealed by spatially filtered magnetoencephalography.
Clin. Neurophysiol. 120 (3), 497–504.
Jensen, O., Tesche, C.D., 2002. Frontal theta activity in humans increases with memory
load in a working memory task. Eur. J. Neurosci. 15 (8), 1395–1399.
Johnson, B.W., Crain, S., Thornton, R., Tesan, G., Reid, M., 2010. Measurement of brain
function in pre-school children using a custom sized whole-head MEG sensor
array. Clin. Neurophysiol. 121 (3), 340–349.
Jokisch, D., Jensen, O., 2007. Modulation of gamma and alpha activity during a working
memory task engaging the dorsal or ventral stream. J. Neurosci. 27 (12), 3244–3251.
Kahana, M.J., Seelig, D., Madsen, J.R., 2001. Theta returns. Curr. Opin. Neurobiol. 11 (6),
739–744.
Karrasch, M., Laine, M., Rapinoja, P., Krause, C.M., 2004. Effects of normal aging on
event-related desynchronization/synchronization during a memory task in
humans. Neurosci. Lett. 366 (1), 18–23.
Karrasch, M., Laine, M., Rinne, J.O., Rapinoja, P., Sinervä, E., Krause, C.M., 2006. Brain os-
cillatory responses to an auditory-verbal working memory task in mild cognitive
impairment and Alzheimer's disease. Int. J. Psychophysiol. 59 (2), 168–178.
Kogure, D., Matsuda, H., Ohnishi, T., Asada, T., Uno, M., Kunihiro, T., Nakano, S.,
Takasaki, M., 2002. Longitudinal evaluation of early Alzheimer's disease using
brain perfusion SPECT. J. Nucl. Med. 41 (7), 1155–1162.
Kurimoto, R., Ishii, R., Canuet, L., Ikezawa, K., Azechi, M., Iwase, M., Yoshida, T., Kazui, H.,
Yoshimine, T., Takeda, M., 2008. Event-related synchronization of alpha activity in
early Alzheimer's disease and mild cognitive impairment: an MEG study combin-
ing beamformer and group comparison. Neurosci. Lett. 443 (2), 86–89.
Li, S.C., Linderberger, U., Sikstrom, S., 2001. Aging condition: from neuromodulation to
representation. Trends Cogn. Sci. 5 (11), 479–486.
Maestú, F., Fernández, A., Simos, P.G., Gil-Gregorio, P., Amo, C., Rodriguez, R., Arrazola,
J., Ortiz, T., 2001. Spatio-temporal patterns of brain magnetic activity during a
memory task in Alzheimer's disease. Neuroreport 12 (18), 3917–3922.
Maestú, F., Campo, P., Gil-Gregorio, P., Fernández, S., Fernández, A., Ortiz, T., 2006.
Medial temporal lobe neuromagnetic hypoactivation and risk for developing cog-
nitive decline in elderly population: a 2-year follow-up study. Neurobiol. Aging
27 (1), 32–37.
Maestú, F., Campo, P., Del Río, D., Moratti, S., Gil-Gregorio, P., Fernández, A., Capilla, A.,
Ortiz, T., 2008. Increased biomagnetic activity in the ventral pathway in mild cog-
nitive impairment. Clin. Neurophysiol. 119 (6), 1320–1327.
Maestú, F., Baykova, E., Ruiz, J.M., Montejo, P., Montenegro, M., Llanero, M., Solesio, E.,
Gil-Gregorio, P., Yubero, R., Paul, N., Pozo, F., Nevado, A., 2011. Increased biomag-
netic activity in healthy elderly with subjective memory complaints. Clin.
Neurophysiol. 123 (3), 499–505.
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., Stadlan, E.M., 1984.
Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work
Group under the auspices of Department of Health and Human Services Task
Force on Alzheimer's Disease. Neurology 34 (7), 939–944.
Missonnier, P., Deiber, M.P., Gold, G., Herrmann, F.R., Millet, P., Michon, A., Fazio-Costa,
L., Ibañez, V., Giannakopoulos, P., 2007. Working memory load-related electroen-
cephalographic parameters can differentiate progressive from stable mild cogni-
tive impairment. Neuroscience 150 (2), 346–356.
Montez, T., Poli, S.S., Jones, B.F., Manshanden, I., Verbunt, J.P., van Dijk, B.W.,
Brussaard, A.B., van Ooyen, A., Stam, C.J., Scheltens, P., Linkenkaer-Hansen,
K., 2009. Alterd temporal correlations in parietal alpha and prefrontal theta
oscillations in early-stage Alzheimer disease. Proc. Natl. Acad. Sci. 106 (5),
1614–1619.
Morris, J.C., 1993. The Clinical Dementia Rating (CDR): current version and scoring
rules. Neurology 43 (11), 2412–2414.
Oshino, S., Kato, A., Wakayama, A., Taniguchi, M., Hirata, M., Yoshimine, T., 2007.
Magnetoencephalographic analysis of cortical oscillatory activity in patients with
brain tumors: synthetic aperture magnetometry (SAM) functional imaging of
delta band activity. Neuroimage 34 (3), 957–964.
Pesonen, M., Björnberg, C.H., Hämäläinen, H., Krause, C.M., 2006. Brain oscillatory 1–30
Hz EEG ERD/ERS responses during the different stages of an auditory memory
search task. Neurosci. Lett. 399 (1–2), 45–50.
Pesonen, M., Hämäläinen, H., Krause, C.M., 2007. Brain oscillatory 4–30 Hz responses
during a visual n-back memory task with varying memory load. Brain Res. 1138,
171–177.
Petersen, R.C., 2004. Mild cognitive impairment as a diagnostic entity. J. Intern. Med.
256 (3), 183–194.
Pfurtscheller, G., Klimesch, W., 1991. Event-related desynchronization during motor
behaviorandvisualinformation
Neurophysiol. Suppl. 42, 58–65.
Pfurtscheller, G., Lopes da Silva, F.H., 1999. Event-related EEG/MEG synchronization
and desynchronization: basic principles (review). Clin. Neurophysiol. 110 (11),
1842–1857.
Ramaroson, H., Helmer, C., Barberger-Gateau, P., Letenneur, L., Dartigues, J.F., 2003.
Prevalence of dementia and Alzheimer's disease among subjects aged 75 years
or over: updated results of the PAQUID cohort. Rev. Neurol. (Paris) 159 (4),
405–411.
Robinson, S.E., Rose, D.F., 1992. Current source image estimation by spatiialy filtered
MEG. In: Hoke, M., Eme, S.E., Okada, Y.C. (Eds.), Biomagnetism: Clinical Aspects:
Proceedings of the 8th International Conference on Biomagnetism. Elsevier, New
York, pp. 761–765.
Stam, C.J., 2000. Brain dynamics in theta and alpha frequency bands and working mem-
ory performance in humans. Neurosci. Lett. 286 (2), 115–118.
Sternberg, S., 1966. High-speed scanning in human memory. Science 153 (736),
652–654.
Uhlhaas, P.J., Singer, W., 2006. Neural synchrony in brain disorders: relevance for cog-
nitive dysfunctions and pathophysiology. Neuron 52 (1), 155–168.
processing.Electroencephalogr.Clin.
4140
R. Kurimoto et al. / NeuroImage 59 (2012) 4132–4140