Available via license: CC BY
Content may be subject to copyright.
Frontiers in Aging Neuroscience 01 frontiersin.org
Oscillatory characteristics of
resting-state
magnetoencephalography reflect
pathological and symptomatic
conditions of cognitive
impairment
HideyukiHoshi
1
*, YokoHirata
2, KeisukeFukasawa
3,
MomokoKobayashi
4 and YoshihitoShigihara
1,4
1 Precision Medicine Centre, Hokuto Hospital, Obihiro, Japan, 2 Department of Neurosurgery,
Kumagaya General Hospital, Kumagaya, Japan, 3 Clinical Laboratory, Kumagaya General Hospital,
Kumagaya, Japan, 4 Precision Medicine Centre, Kumagaya General Hospital, Kumagaya, Japan
Background: Dementia and mild cognitive impairment are characterised
by symptoms of cognitive decline, which are typically assessed using
neuropsychological assessments (NPAs), such as the Mini-Mental
State Examination (MMSE) and Frontal Assessment Battery (FAB).
Magnetoencephalography (MEG) is a novel clinical assessment technique that
measures brain activities (summarised as oscillatory parameters), which are
associated with symptoms of cognitive impairment. However, the relevance of
MEG and regional cerebral blood flow (rCBF) data obtained using single-photon
emission computed tomography (SPECT) has not been examined using clinical
datasets. Therefore, this study aimed to investigate the relationships among
MEG oscillatory parameters, clinically validated biomarkers computed from
rCBF, and NPAs using outpatient data retrieved from hospital records.
Methods: Clinical data from 64 individuals with mixed pathological backgrounds
were retrieved and analysed. MEG oscillatory parameters, including relative
power (RP) from delta to high gamma bands, mean frequency, individual alpha
frequency, and Shannon’s spectral entropy, were computed for each cortical
region. For SPECT data, three pathological parameters—‘severity’, ‘extent’, and
‘ratio’—were computed using an easy z-score imaging system (eZIS). As for
NPAs, the MMSE and FAB scores were retrieved.
Results: MEG oscillatory parameters were correlated with eZIS parameters. The
eZIS parameters associated with Alzheimer’s disease pathology were reflected
in theta power augmentation and slower shift of the alpha peak. Moreover, MEG
oscillatory parameters were found to reflect NPAs. Global slowing and loss of
diversity in neural oscillatory components correlated with MMSE and FAB scores,
whereas the associations between eZIS parameters and NPAs were sparse.
Conclusion: MEG oscillatory parameters correlated with both SPECT (i.e.
eZIS) parameters and NPAs, supporting the clinical validity of MEG oscillatory
parameters as pathological and symptomatic indicators. The findings indicate
that various components of MEG oscillatory characteristics can provide valuable
pathological and symptomatic information, making MEG data a rich resource
for clinical examinations of patients with cognitive impairments. SPECT (i.e.
eZIS) parameters showed no correlations with NPAs. The results contributed
OPEN ACCESS
EDITED BY
Yang Jiang,
University of Kentucky, UnitedStates
REVIEWED BY
Diego Pinal,
University of Santiago de Compostela, Spain
Tie-Qiang Li,
Karolinska University Hospital, Sweden
Mate Gyurkovics,
University of Illinois at Urbana-Champaign,
UnitedStates
*CORRESPONDENCE
Hideyuki Hoshi
heurekaesthem.avir@gmail.com
RECEIVED 07 August 2023
ACCEPTED 12 January 2024
PUBLISHED 30 January 2024
CITATION
Hoshi H, Hirata Y, Fukasawa K,
Kobayashi M and Shigihara Y (2024)
Oscillatory characteristics of resting-state
magnetoencephalography reflect
pathological and symptomatic conditions of
cognitive impairment.
Front. Aging Neurosci. 16:1273738.
doi: 10.3389/fnagi.2024.1273738
COPYRIGHT
© 2024 Hoshi, Hirata, Fukasawa, Kobayashi
and Shigihara. This is an open-access article
distributed under the terms of the Creative
Commons Attribution License (CC BY). The
use, distribution or reproduction in other
forums is permitted, provided the original
author(s) and the copyright owner(s) are
credited and that the original publication in
this journal is cited, in accordance with
accepted academic practice. No use,
distribution or reproduction is permitted
which does not comply with these terms.
TYPE Original Research
PUBLISHED 30 January 2024
DOI 10.3389/fnagi.2024.1273738
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 02 frontiersin.org
to a better understanding of the characteristics of electrophysiological and
pathological examinations for patients with cognitive impairments, which will
help to facilitate their co-use in clinical application, thereby improving patient
care.
KEYWORDS
magnetoencephalography, cognitive impairments, clinical neurophysiology,
neurology, single-photon emission computed tomography
1 Introduction
Dementia and related cognitive impairments, such as mild
cognitive impairment (MCI), remain major challenges in medicine
due to neurological diseases such as Alzheimer’s disease (AD),
neurovascular diseases (e.g. stroke and cerebral hemorrhage), and
hydrocephalus (Elahi and Miller, 2017). Cognitive impairments and
neurological diseases have a cause-eect relationship, where cognitive
impairments (i.e. symptoms) are attributed to neurological diseases
(i.e. pathologies). However, this relationship is nonlinear; for example,
patients with AD do not consistently exhibit cognitive impairment
(Snowdon, 1997; Shiroky et al., 2007). erefore, these distinct
concepts should beevaluated separately. Assessing the severity and
subtyping of cognitive impairments (i.e. symptoms) is essential in the
clinical treatment of dementia and MCI given that these diseases are
dened by their symptoms (American Psychiatric Association, 2013;
Sachdev etal., 2014), not causative pathologies, and are primarily
assessed using neuropsychological assessments (NPAs) such as the
Mini-Mental State Examination (MMSE) (Folstein et al., 1975;
Shigemori etal., 2010) and Frontal Assessment Battery (FAB) (Dubois
etal., 2000). Although these assessments are well-established and
validated, they have inherent drawbacks, such as objectivity, test
security, and dependency on the examinees’ background and
experience (Heilbronner, 2007). In particular, practice eects interfere
with the accuracy of long-term monitoring and follow-up of patients’
neurocognitive status (Wahlstrom and Boersma, 1968; Galasko etal.,
1993; McCarey and Westervelt, 1995; Calamia etal., 2012).
Magnetoencephalography (MEG) and electroencephalography
(EEG) are new clinical examination tools for supporting NPAs
(Vecchio etal., 2013; Engels etal., 2017; Malek etal., 2017; López-
Sanz etal., 2018; Mandal etal., 2018; Maestú etal., 2019; Giustiniani
etal., 2023). MEG measures the magnetic elds generated by cortical
neuronal/synaptic activity, a direct indication of brain activity (Hari
and Puce, 2017; Hari etal., 2018). Patients with cognitive impairment
show unique brain activity, including changes in oscillatory
characteristics, based on resting-state MEG data. e MEG
oscillatory characteristics of patients with cognitive impairment
include (i) enhanced low-frequency oscillatory activity accompanied
by attenuated high-frequency oscillatory activity, (ii) slowing down
of the alpha peak frequency, (iii) less prominent alpha oscillations,
and (iv) loss of diversity of neural oscillatory components (Poza
etal., 2007; Fernández etal., 2013; López etal., 2014). ese features
are summarised as clinical parameters, such as mean frequency
(MF), individual alpha frequency (IAF), Shannon’s spectral entropy
(SSE), and relative power (RP), which have been used in clinical
practice at our memory clinics (Hoshi etal., 2022; Hirata et al.,
2024). Changes in MEG oscillatory parameters serve as direct
measures of modications in the neuronal or synaptic activities of
the brain and are, therefore, strongly associated with cognitive
impairment (i.e. cognitive symptoms) (Hoshi et al., 2022).
Furthermore, MEG oscillatory parameters are advantageous in their
clinical use as they require low computational costs (e.g. the
calculation requires approximately 10 min using a laptop computer).
Single-photon emission computed tomography (SPECT) has been
used in clinical practice for more than 20 years (Jagust etal., 1995;
Davis etal., 2020) to assess the level of perfusion by measuring the
regional cerebral blood ow (rCBF), which is aected by various
pathological conditions, such as AD (Albert etal., 2011) and dementia
with Lewy bodies (DLB) (McKeith etal., 2017). Analysis pipelines,
such as the ‘easy Z-score imaging system (eZIS)’, are well established.
e eZIS parameters are a set of biomarkers that indicate the
probability of AD (Matsuda etal., 2007) or DLB (Imabayashi etal.,
2017) and parameterise changes in the rCBF, which are aected by the
pathological conditions of these diseases. e levels of hypoperfusion
and hypometabolism are midstream pathological biomarkers of AD
(Bjorkli etal., 2020), reecting the synaptic dysfunction caused by
these diseases (Fagan, 2014). Based on the assumption of
neurovascular coupling (Iadecola, 2017), rCBF is considered an
indirect measurement of neural activity. erefore, there is an
association between MEG/EEG oscillatory parameters (i.e. direct
measurements of synaptic dysfunction) and eZIS parameters (i.e.
measurements of rCBF/hypoperfusion and clinically validated
midstream biomarkers). No prior research has explored the
correlation between MEG oscillatory and eZIS parameters. When
incorporating MEG and eZIS parameters into clinical practice, it is
crucial to accurately grasp both their commonalities and distinctions.
erefore, in this study, weinvestigated the relationship between
MEG oscillatory and eZIS parameters using actual clinical data
obtained retrospectively from the clinical records of individuals who
visited the outpatient department of dementia at our hospital.
Additionally, to examine the relevance to clinical symptoms, two NPA
scores, the MMSE and FAB, were retrieved from the records, and their
relationships with MEG oscillatory and eZIS parameters were
assessed. We hypothesise that MEG oscillatory parameters are
associated with both pathological and symptomatic changes and that
their symptomatic associations are stronger than those of the eZIS
parameters since eZIS parameters are indirect measurements of neural
activity (i.e. neurovascular coupling). According to this hypothesis,
weexpected that MEG oscillatory parameters would becorrelated
with both eZIS parameters and NPAs, whereas the direct relationships
between eZIS parameters and NPAs would be sparse. A better
understanding of the similarities and dierences between MEG
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 03 frontiersin.org
oscillatory and eZIS parameters will help to facilitate their clinical
application, thereby improving patient care.
2 Materials and methods
2.1 Participants and ethics
e clinical records of 72 individuals who visited the outpatient
department of Kumagaya General Hospital (Saitama, Japan) were
retrieved. Weset rough exclusion criteria to align the background of
individuals in this study as closely as possible with that of the clinical
population in the outpatient department. is led to the exclusion of
(1) one individual with a non-dementia disease, (2) four individuals
who were undiagnosed at the time of examination, and (3) three
individuals whose MEG data were contaminated from severe artifacts.
Consequently, data from the remaining 64 individuals were included
in the analysis (35 females, mean age ± standard deviation:
77.0 ± 7.0 years; age range: 53–91 years; Table1). e retrieved data
included MEG data, eZIS parameters computed from the SPECT
measurements, and NPAs. Among the 64 individuals, 9 had healthy
ageing, 15 were diagnosed with MCI, 30 with AD, and 10 with other
types of dementia [i.e. vascular dementia (VD), DLB, frontotemporal
dementia (FTD), or a combination of them]. Detailed information on
each diagnostic group is shown in Supplementary Table S1. is study
was conducted in accordance with the ethical principles of the
Declaration of Helsinki and was approved by the Ethics Committee of
Kumagaya General Hospital (approval number: #76). All individuals
provided written informed consent to participate in this study if they
were cognitively healthy. Otherwise, their legal guardians/next of kin
provided consent on their behalf.
2.2 Neuropsychological assessments
For the NPAs, scores from two assessments, namely MMSE and
FAB, were retrieved. ese assessments are scored on scales of 0–30
and 0–18, respectively, where lower scores indicate more severe
cognitive impairment. MMSE is the most commonly used assessment
tool for dementia screening (Shigemori etal., 2010) and primarily
evaluates learning/memory performance (Dinomais etal., 2016). In
contrast, FAB assesses executive functions, including attention,
inhibition, working memory, interference control, and cognitive
exibility (Diamond, 2013), which are served by the prefrontal cortex
(Gilbert and Burgess, 2008). e assessments were administered by
clinical psychologists as part of clinical practice. Notably, FAB was not
administered to an individual who did not wish to beassessed; thus,
all statistical analyses involving FAB data were performed using data
from 63 individuals (out of 64 individuals). e descriptive statistics
of the NPAs are summarised in Table1.
2.3 SPECT measurement and data analysis
To assess the pathological conditions, weused three parameters—
namely ‘severity’, ‘ extent’, and ‘ratio’—determined using the eZIS
soware (PDRadiopharma Inc., Tokyo, Japan). ese parameters were
based on rCBF images obtained through SPECT (Mizumura and
Kumita, 2006; Matsuda etal., 2007), which are sensitive to pathological
changes in conditions such as AD. SPECT scanning was performed as
part of the clinical assessment, and the scores were retrieved from the
clinical records. Prior to SPECT measurement, all participants
received an intravenous injection of 600 MBq of 99mTc-ethylcysteinate
dimer (PDRadiopharma Inc. Tokyo, Japan). Aer 5 min of rest with
eyes closed in the supine position in a dark room, SPECT data were
obtained using a 128 × 128 matrix on a BrightView X (Philips
Healthcare, Milpitas, CA, UnitedStates) equipped with a low-energy,
cardiac high-resolution parallel-hole collimator and dual thallium-
activated sodium iodide scintillation detector. Seventy-two views were
obtained continuously throughout the 360° rotation (5°/step, zoom
1.85) with a pixel size of 3.2 mm. e acquired images were
reconstructed using the ltered back-projection method with
combined Chang attenuation correction. e reconstructed rCBF
images were analysed using eZIS, which implemented the Statistical
Parametric Mapping 2 toolbox (Wellcome Trust Centre for
Neuroimaging, London, UK
1
) and automated image processing,
including spatial normalisation, smoothing, and Z-scoring, by
comparing the data against an age-matched control database. In the
eZIS soware, the Z-score was obtained as [(control mean) –
(individual value)]/(control SD), where a higher Z-score indicates
more reduction in rCBF and severe hypoperfusion (Matsuda etal.,
2007). e Z-scored images were summarised within a pre-dened
1 https://www.fil.ion.ucl.ac.uk/spm/
TABLE1 Descriptive statistics.
MSD MIN MAX
Age 76.95 6.977 53 91
NPA s MMSE 24.16 5.109 9 30
FAB 11.97 3.137 4 18
eZIS severity 1.299 0.495 0.680 3.170
extent 15.84 14.80 0.760 64.62
ratio 2.109 1.615 0.060 6.130
MEG RPd 0.192 0.067 0.064 0.339
RPt 0.124 0.045 0.045 0.280
RPa1 0.112 0.043 0.034 0.231
RPa2 0.110 0.046 0.044 0.286
RPa3 0.063 0.021 0.031 0.141
RPb 0.218 0.068 0.073 0.450
RPlg 0.068 0.022 0.022 0.133
RPhg 0.052 0.019 0.018 0.108
MF 9.425 2.120 5.650 16.68
IAF 8.665 0.800 6.865 10.38
SSE 0.787 0.029 0.714 0.840
M, mean; SD, standard deviation; NPAs, neuropsychological assessments; MMSE, Mini-
Mental State Examination; FAB, Frontal Assessment Battery; eZIS, easy Z-score imaging
system; MEG, magnetoencephalography; RPd, relative power in delta band; RPt, relative
power in theta band; RPa1, relative power in alpha1 band; RPa2, relative power in alpha2
band; RPa3, relative power in alpha3 band; RPb, relative power in beta band; RPlg, relative
power in low gamma band; RPhg, relative power in high gamma band; MF, mean frequency;
IAF, individual alpha frequency; SSE, Shannon’s spectral entropy.
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 04 frontiersin.org
volume of interest (VOI) set bilaterally on the posterior cingulate
gyrus (PCG), precuneus (PC), and parietal lobe. is selection was
based on a previous study that demonstrated group-level dierences
in rCBF images between patients with amnestic MCI (aMCI) due to
AD and individuals with healthy ageing (Matsuda et al., 2007).
Notably, although the cortical regions (PCG, PC, and parietal lobe)
were spatially discontinuous, they were considered a single VOI. ree
primary parameters were computed using the Z-score summarised in
the VOI: ‘severity,’ ‘ extent’, and ‘ratio’ (Matsuda etal., 2007; Hayashi
et al., 2020). e severity indicated the degree of decrease in the
average Z-scored rCBF in the VOI. e extent indicated the percentage
of voxels with a Z-score > 2 relative to the total number of voxels in the
VOI. e ratio indicated the specicity of rCBF reduction in the VOI,
which is the rate of the extent value to that computed using the whole-
brain volume instead of the VOI. For the three parameters, higher
values corresponded to more severe hypoperfusion of the VOI. To
discriminate patients with aMCI due to AD from individuals with
healthy ageing, optimal cut-o thresholds were set at 1.19, 14.2, and
2.22 for severity, extent, and ratio, respectively (Matsuda etal., 2007).
e descriptive statistics of the eZIS parameters are summarised in
Table1.
2.4 MEG measurement
Resting-state MEG data were collected from clinical records. e
resting-state cortical activity was recorded for 5 min using a whole-
head-type MEG system (RICOH160-1; RICOH Co. Ltd., Tokyo, Japan)
equipped with 160-channel axial gradiometers and placed in a
magnetically shielded room at Kumagaya General Hospital. During the
scan, the participants were asked to remain awake and calm in a supine
position with their eyes closed. e sensor coils were gradiometers with
diameters of 15 mm and heights of 50 mm. e pairs of sensor coils
were separated 23 mm apart. e sampling frequency was 2,000 Hz
with 500-Hz low-pass ltering during recording. To co-register the
MEG data with the anatomical brain images, ve ducial magnetic
marker coils were placed on each participant’s face (40 mm above the
nasion, bilaterally 10 mm in front of the tragus, and at the bilateral
pre-auricular points) prior to the MEG scan. e spatial coordinates of
these markers were measured immediately before scanning. To
maintain the optimal state of vigilance, the recording was initiated
shortly aer closing the door of the magnetically shielded room.
Participants were also reminded to stay awake with their eyes closed
via the intercom prior to scanning, if necessary. During the scan,
participants were monitored by medical technicians using a video
camera installed in a magnetically shielded room, and their vigilance
states were veried using self-reports collected aer the scan.
2.5 MEG data analysis
MEG data were analysed oine using the RICOH MEG Analysis
soware (RICOH, Tokyo, Japan), MATLAB (MathWorks, MA,
UnitedStates), and Brainstorm, which is documented and freely
available for download online under the GNU general public license2
2 http://neuroimage.usc.edu/brainstorm
(Tadel etal., 2011). e pre-processing pipeline followed the strategy
used in a previous study (Rodríguez-González etal., 2020). First,
continuous MEG signals were cleaned using a dual-signal subspace
projection algorithm (Sekihara etal., 2016) available on vendor-
provided soware (RICOH MEG Analysis), which is comparable to
the temporally extended signal space separation algorithm, with the
only dierence being the approximation of the signal subspace
projector (Cai etal., 2019). Next, to remove the remaining artifacts,
the signals were decomposed with independent component analysis
(ICA) using the FastICA algorithm implemented in Brainstorm
(Makeig etal., 1996). Each ICA component was visually inspected,
and those for cardiac, blinking, and other salient artifacts were
rejected. Artifact-free signals were ltered using Finite Impulse
Response Filtering with a Hamming window by applying a bandpass
(1–70 Hz) to limit the noise bandwidth and a bandstop (48–52 Hz)
to remove line noise.
e ltered signals were then imported to Brainstorm, where
they were projected onto the cortical source using the soware’s
default parameters. ICBM152, a template anatomical brain image
prepared by Brainstorm, was used for the analysis. ICBM152 is a
nonlinear average of 152 magnetic resonance images from dierent
subjects (Fonov etal., 2009) and is provided along with its cortical
segments. e signal source was restricted to the cortex, which was
segmented into 15,000 vertices. Each MEG data set was co-registered
with the anatomical image using the spatial coordinates of ve
ducial points and the nasion, and the relationship between 160
MEG channels and 15,000 vertices (i.e. leadeld matrix) was
modeled (i.e. forward modeling) using a Symmetric Boundary
Element Method (Kybic etal., 2005; Gramfort etal., 2010), which
generated a three-layer (brain, skull, and scalp) realistic head model.
Prior to computing the source signals, the characteristics of the
MEG sensor noise were modeled as a covariance matrix for each pair
of channels (i.e. noise covariance), which was dened as the average
of the covariance matrices across eight empty-room recordings
(5 min) measured using an identical MEG machine with the same
acquisition setting. Using the forward model and noise covariance
matrix, the source signals of MEG data were computed using the
Weighted Minimum Norm Estimation (wMNE) method (Lin etal.,
2006). e wMNE restricts the sources of the inverse problem by
minimising the energy (L2 norm) while weighting the deep sources
to facilitate their detection. is algorithm was selected for three
reasons: (i) it is recommended as a default option in Brainstorm, (ii)
weutilised template brain instead of individual MRI images, which
provides only rough approximations of the forward model and is
unsuitable for other inversion algorithms (e.g. Beamforming)
requiring better model approximation than wMNE, and (iii) wMNE
is widely used in the context of MEG and EEG source localization
for studying oscillatory characteristics of pathological conditions
(Larson-Prior etal., 2013; Rizkallah etal., 2020; Rodríguez-González
etal., 2020; Tait etal., 2021). e orientation of the neural sources
was restricted to benormal to the cortex. e resulting data from the
source reconstruction process were continuous time-series signals
for each of the 15,000 cortical vertices. High-dimensional data were
limited to 103 anatomical regions dened by the Automated
Anatomical Labeling Atlas 3 (AAL3) (Rolls etal., 2020) by averaging
the signals of the vertices included in each anatomical region.
Weused the AAL3 atlas because it corresponds to the atlas used for
labeling the VOI in eZIS (Imabayashi etal., 2016, 2017). During the
averaging process, the signs of the signals were ipped in the
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 05 frontiersin.org
vertices, where the normal orientation was opposite to the dominant
orientation of the corresponding region.
To extract the oscillatory power from regional time-series signals,
the power spectral density (PSD) was computed using the Blackman-
Tukey approach (Blackman and Tukey, 1958) with non-overlapping
5-s segments. In the Blackman-Tukey method, PSD is dened as a
discrete Fourier transform of the autocorrelation function of the time-
series data, which has better precision than other approaches (Kale
etal., 2013) and is commonly used for computing MEG oscillatory
parameters (Poza etal., 2007, 2008a,b; Gómez etal., 2013; Rodríguez-
González etal., 2020; Hoshi etal., 2022). To obtain the normalised
PSD (PSDn), the original PSD was divided by the total power in the
frequency range of interest, i.e. 1–70 Hz (Gómez etal., 2013). Next, the
RP in each canonical frequency band [delta (1–3 Hz), theta (4–7 Hz),
alpha1 (7–9 Hz), alpha2 (9–11 Hz), alpha3 (11–13 Hz), beta
(13–25 Hz), and gamma (low gamma, 26–40 Hz; high gamma,
41–70 Hz)] was computed by calculating the cumulative sum of the
power. e RP in each band was referred to as RPd (delta), RPt (theta),
RPa1 (alpha1), RPa2 (alpha2), RPa3 (alpha3), RPb (beta), RPlg (low
gamma), and RPhg (high gamma). Notably, the alpha band was
subdivided into three sub-bands (alpha1, alpha2, and alpha3) because
their correlational behaviours to the level of cognitive impairments
were expected to beopposite before and aer the alpha peak (Poza
etal., 2007; Gómez etal., 2013; López etal., 2014), which would cancel
out if the bands were considered as whole. Previous studies proposed
exible and individually adjusted denitions of frequency bands
(Babiloni etal., 2020), including three alpha sub-bands (Klimesch,
1999; Doppelmayr etal., 2002); however, this study focused on the
existing MEG oscillatory parameters, which are dened using xed
canonical frequency bands. Hence, the canonical alpha sub-bands
were set following a previous study (Wu and Liu, 1995). Finally, three
spectral parameters were calculated to summarise the dierent
properties of PSDn: MF, IAF, and SSE (Poza et al., 2008b), the
denitions and details of which have been previously reported (Poza
etal., 2007). e rst parameter, MF, quanties the frequency at which
the spectral power is balanced between low and high frequencies. e
frequency divides PSDn into two halves, 1 Hz and 70 Hz. e second
parameter, IAF, represents the dominant frequency corresponding to
the peak of the PSDn in the alpha band and is dened similarly to MF,
except that the frequency range is adjusted to 4–15 Hz (i.e. extended
alpha band), to obtain a robust estimator of the dominant alpha
oscillations (Poza etal., 2007). e last parameter, SSE, is dened by
applying the denition of the normalised Shannon entropy to PSDn,
which can be assimilated as a probability density function (Poza
etal., 2007):
S
SE
1
1
70
log
·l
og
NPSDn fPSDnf
fHz
Hz
Where N is the number of frequency bins of PSDn. SSE represents
an irregularity measure closely related to the concept of order in
information theory, which quanties the homogeneity in the
distribution of the oscillatory components of the PSDn. All spectral
parameters (MF, IAF, and SSE) have been considered as reections of
the levels of cognitive impairments (Engels etal., 2017; López-Sanz
etal., 2018; Mandal etal., 2018; Maestú etal., 2019), presumably
capturing dierent aspects of the cognitive impairments. is
interpretation is supported by a previous study that demonstrated the
relationships between MF and MMSE/FAB, IAF and MMSE, and SSE
and FAB (Hoshi et al., 2022). erefore, in this study, the three
parameters were considered separately. Region- and epoch-wise RPs,
MF, IAF, and SSE, were computed, averaged across epochs, and used
in the statistical analysis. Descriptive statistics of the average MEG
oscillatory parameters are summarised in Table 1. e regional
distribution of the MEG oscillatory parameters is shown in
Supplementary Figure S1.
2.6 Statistical analysis
Statistical analyses were performed using MATLAB (MathWorks,
MA, UnitedStates), the Fieldtrip toolbox (Maris and Oostenveld,
2007; Oostenveld etal., 2011), the Statistics and Machine Learning
Toolbox (MathWorks, Natick, MA, UnitedStates), and the Multiple
Testing Toolbox (Martínez-Cagigal, 2021).
In previous studies, the MEG oscillatory parameters were rst
examined by summarising across all sensors (Fernández etal., 2006;
Poza etal., 2007, 2008a,b; Gómez etal., 2013), and their global changes
were considered as clinical parameters for capturing cognitive
impairments (Hoshi et al., 2022). erefore, rst, the global
relationships between MEG oscillatory parameters (RPs in eight
canonical frequency bands, MF, IAF, and SSE) averaged across
anatomical regions, three eZIS parameters (severity, extent, and ratio),
two NPA scores (MMSE and FAB), and the age of the individuals were
assessed. For each pair of parameters, a 95% bootstrap condence
interval of Spearman’s coecient (rho) was computed by 5,000
bootstrap iterations, and the correlation was considered signicant
when the interval did not contain 0. e within-modality correlations
among MEG oscillatory or eZIS parameters (e.g. RPd × RPt,
severity × extent) were not examined because they are not within the
scope of this study. Moreover, two biasing factors exist for the within-
modality correlations of MEG oscillatory parameters: 1/f activity and
normalisation. e changes in slope (i.e. steepness) of 1/f activity, a
broadband pattern underlying the PSDn, lead to rotational reshaping
of PSDn, resulting in antagonistic behaviours between low- and high-
frequency bands. Additionally, changes in intercept (i.e. oset)
contribute to overall increasing/decreasing of the PSDn, causing
sympathetic behaviours between low- and high-frequency bands. e
normalisation converts the absolute power to relative power across
frequency bins, creating a pseudo trade-o relationship between
them. Due to their intercorrelating nature in the denition, wedeemed
evaluations of within-modality correlations not meaningful. Notably,
the nonlinear Spearman’s correlation is robust for the outliers (de
Winter etal., 2016). erefore, wedid not screen out the outliers in
the MEG oscillatory parameters but instead removed the dataset by
inspecting the raw MEG signals (see Section 2.1).
Next, the regional relationships between the MEG oscillatory
parameters, three eZIS parameters, and two NPA scores were
examined using a non-parametric cluster-based permutation
approach implemented in the Fieldtrip toolbox (Maris and
Oostenveld, 2007). e non-parametric cluster-based permutation
approach was used for controlling family-wise error rate in multiple
comparisons. is approach empirically identies ‘clusters’,
encompassing multiple adjacent regions where the neuroimaging data
exhibit similar behaviours concerning the eect of interest (T-statistic).
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 06 frontiersin.org
As a rst step of the permutation test, for each pair of MEG oscillatory
parameters and eZIS parameters or NPA scores, the T-statistic of
Spearman’s coecient (rho) was calculated in the region (103) space
(observed statistics) using a function implemented in the Fieldtrip
toolbox (_statfun_correlationT.m) with a following formula:
T
rhoN
rho
2
12
Where N is the number of observations (individuals). The
T-statistic was also computed using a shuffled dataset across
participants (e.g. the MEG parameter for participant A was
paired with the eZIS parameter of participant B) 5,000 times,
which generated a probability distribution of the random
statistics. For each region, the observed statistics were examined
if they were above or below the critical value (0.025) in the right
or left tail of the random probability distribution, respectively.
The regions where the observed statistics exceeded the tails were
clustered according to their spatial adjacency on each side of the
tail and defined as positive and negative clusters. The observed
T-statistic was summed in each cluster (observed cluster
statistics), while the random distribution of cluster statistics was
generated by collecting the maximum of summed T-statistics
among detected clusters in the shuffled data for each of the 5,000
iterations (‘maxsum’ method in cluster-based permutation
algorithm in the Fieldtrip toolbox). The percentage of random
cluster statistics that were larger (for positive clusters) or smaller
(for negative clusters) than the observed cluster statistics was
considered as the significance level (p-value) of each observed
cluster. Wereported the average rho [rho (mean)] and average
T-statistics [T (mean)] across regions included in each cluster,
and corresponding p-value. Notably, while the cluster-based
permutation approach addressed the multiple comparison
problem between multiple regions, we did not address the
problem between multiple pairs of the cluster-based permutation
tests (i.e. repeated use of the tests). This decision was based on
the reason that the analyses were exploratory; regarding the MEG
oscillatory and eZIS parameters, lacking a priori hypotheses and
focusing on specific parameters.
The potential confounding effects of age were examined in
the final step of the cluster-based permutation test. As the global
correlation analysis revealed that age correlated with the ratio of
eZIS parameters and the two NPA scores (i.e. MMSE and FAB),
their relationships with MEG oscillatory parameters could
beconfounded by age (i.e. pseudo-correlations). To examine this,
the contributions of the three variables (ratio, MMSE, or FAB)
and age to the MEG oscillatory parameters were evaluated using
the partial least squares (PLS) regression. Taking each MEG
oscillatory parameter in each region as a response variable, PLS
regression was performed using two predictors, namely one of
the three variables and age, with a one-component model. The
variable importance in projection (VIP) scores were computed
for each predictor and compared for each MEG oscillatory
parameter in every region. Larger VIP scores for age compared
to those for one of the other variables (ratio, MMSE, and FAB)
indicated that the contribution of age was stronger to the regional
MEG oscillatory parameter than the variable. In such cases,
weconsidered the possibility of potential confounding effects of
age on the relationship between the variable and the regional
MEG oscillatory parameter. Therefore, the region was excluded
from the significant clusters found in the cluster-based
permutation tests for the variable.
To aid the interpretation of the results, statistical analyses were
repeated using data from patients with MCI and AD (N = 45),
presented in Supplementary sections B–D.
3 Results
3.1 Global correlations between MEG
oscillatory parameters, eZIS parameters,
and NPAs
e results of the global correlations between the MEG oscillatory
parameters, eZIS parameters, NPAs, and age are illustrated in Figure1
(MEG vs. the others), Figure 2 (between the others), and
Supplementary Tables S7, S8.
Figure 1 and Supplementary Table S7 summarise the
between-modality correlations of MEG parameters with eZIS
parameters, NPAs, and individuals’ age. Among the MEG × eZIS
parameters, RPt was positively correlated with severity. For
MEG × NPAs, the RPs in slower bands (RPt and RPa1) showed
constant negative correlations with NPAs, whereas the RPs in
faster bands (RPa3, RPb, and RPlg) showed positive correlations.
Moreover, the MEG spectral parameters also showed positive
correlations with NPAs. For MEG × age, it was only
sparsely correlated with the MEG oscillatory parameters of RPa1,
RPlg, and RPhg. No other pair showed significant
global correlations.
Figure2 and Supplementary Table S8 summarise the results of
between-modality correlations of eZIS parameters, NPAs, and age.
e eZIS parameters did not correlate with NPAs, while the ratio was
correlated negatively with age. e NPAs showed strong positive
intercorrelations between MMSE and FAB and negative correlations
with age.
3.2 Regional correlations between MEG
oscillatory and eZIS parameters
Global correlation analysis demonstrated that the average MEG
oscillatory parameters across all regions were associated with the
eZIS parameters. To investigate the cortical regions involved in these
associations, weconducted regional analyses of the relationships
using cluster-based permutation tests in the region (103) space
(Figure 3). RPt showed positive correlations with severity [rho
(mean) = 0.281, T (mean) = 2.310, p = 0.012] in globally distributed
regions, particularly prominent in the right hemisphere (Figure3A).
Similarly, IAF demonstrated negative correlations with severity [rho
(mean) = −0.279, T (mean) = −2.290, p = 0.022] in globally
distributed regions, which was also more prominent in the right
hemisphere (Figure3B). e regions included in the clusters are
listed in Supplementary Table S9. No signicant clusters were
observed for the correlations between other pairs of MEG oscillatory
parameters and eZIS parameters.
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 07 frontiersin.org
3.3 Regional correlations between MEG
oscillatory parameters and NPAs
Global correlation analyses demonstrated that the average
MEG oscillatory parameters across all regions were associated
with NPAs. To investigate the specific cortical regions involved
in the associations, we conducted analyses of regional
relationships using cluster-based permutation tests in the region
(103) space (Figures4, 5). MMSE (Figure4) showed negative
correlations with RPs in slower bands; global RPt [rho
(mean) = −0.387, T (mean) = −3.332, p < 0.001] and dorsal RPa1
[rho (mean) = −0.374, T (mean) = −3.188, p = 0.002]
(Figures 4A,B), positive correlations with faster bands;
occipitotemporal RPa3 [rho (mean) = 0.378, T (mean) = 3.248,
p = 0.003], global RPb [rho (mean) = 0.373, T (mean) = 3.204,
p = 0.001], and dorsal RPlg [rho (mean) = 0.365, T (me an) = 3.119,
p = 0.003] (Figures4C–E), and positive correlations with spectral
parameters; global MF [rho (mean) = 0.373, T (mean) = 3.214,
p = 0.001], IAF [rho (mean) = 0.423, T (mean) = 3.731, p < 0.001],
and dorsal SSE [rho (mean) = 0.316, T (mean) = 2.641, p = 0.003]
(Figures4F–H).
e FAB had similar regional relationships to MEG oscillatory
parameters as the MMSE (Figure 5). It demonstrated negative
correlations with RPs in slower bands; global RPt [rho (mean) = −0.345,
T (mean) = −2.888, p = 0.003] and dorsal RPa1 [rho (mean) = −0.345,
T (mean) = −2.879, p = 0.009] (Figures 5A,B), positive correlations
with faster bands; occipitotemporal RPa3 [rho (mean) = 0.419, T
(mean) = 3.687, p = 0.002], occipitotemporal RPb [rho (mean) = 0.351,
T (mean) = 2.964, p = 0.001], dorsal RPlg [rho (mean) = 0.371, T
(mean) = 3.161, p = 0.001], and dorsal RPhg [rho (mean) = 0.347, T
(mean) = 2.906, p = 0.010] (Figures5C–F), and positive correlations
with spectral parameters; global MF [rho (mean) = 0.369, T
(mean) = 3.138, p = 0.001], IAF [rho (mean) = 0.379, T (mean) = 3.243,
p < 0.001], and dorsal SSE [rho (mean) = 0.379, T (mean) = 3.242,
p < 0.001] (Figures 5G–I). e regions included in the clusters are
listed in Supplementary Tables S10–S15. No signicant clusters were
observed for the correlations between the other pairs of MEG
oscillatory parameters and NPAs.
FIGURE1
Between-modality correlations between MEG oscillatory parameters averaged across anatomical regions, eZIS parameters, NPAs, and age. The
scatterplot is colored dierently for each diagnosis (black: healthy ageing, blue: MCI, red: AD, and grey: other types of dementia). Regression lines are
added for significant correlations. The number displayed at the corner of each plot indicates Spearman’s coecient (rho) averaged across bootstrap
iterations, with an asterisk (*) indicating significant correlation. eZIS, easy Z-score imaging system; NPAs, neuropsychological assessments; MMSE, Mini-
Mental State Examination; FAB, Frontal Assessment Battery; MEG, magnetoencephalography; RPd, relative power in delta band; RPt, relative power in
theta band; RPa1, relative power in alpha1 band; RPa2, relative power in alpha2 band; RPa3, relative power in alpha3 band; RPb, relative power in beta
band; RPlg, relative power in low gamma band; RPhg, relative power in high gamma band; MF, mean frequency; IAF, individual alpha frequency; SSE,
Shannon’s spectral entropy.
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 08 frontiersin.org
4 Discussion
is study revealed that the MEG oscillatory parameters correlated
with the eZIS parameters. Specically, the clinical parameters related to
AD pathology were linked to the theta power augmentation and slower
shi of the alpha peak, predominantly observed in the right hemisphere.
Furthermore, MEG oscillatory parameters were found to reect NPAs;
global slowing and loss of diversity in the neural oscillatory components
were correlated with MMSE and FAB scores, whereas the associations
between eZIS parameters and NPAs were less pronounced. Overall, the
results largely supported our hypothesis that MEG oscillatory parameters
are associated with both pathological and symptomatic changes and that
their symptomatic relationships are more explicit than those of the
eZIS parameters.
In this study, we focused on the MEG oscillatory parameters
because they are useful in clinical examinations for supporting NPAs
(Engels etal., 2017; López-Sanz et al., 2018; Mandal et al., 2018;
Maestú et al., 2019). Similarly to MEG, EEG also captures
electrophysiological activities, which can besummarise as oscillatory
parameters and shown to be sensitive to cognitive impairment
(Vecchio etal., 2013; Malek etal., 2017; Giustiniani etal., 2023). MEG
and EEG each have their own set of advantages and disadvantages.
From a practical viewpoint, EEG is advantageous in terms of facility
accessibility (i.e. prevalence) and installation/maintenance costs.
However, its preparation requires a lot of eort, incurring costs for
busy clinicians. Moreover, setting up EEG involves attaching
electrodes to the scalp with gel paste, requiring patients to remain
seated stably, which oen is uncomfortable and challenging for
patients with cognitive impairment. On the other hand, while MEG is
expensive and only available in limited hospitals, its preparation and
measurement processes are more straightforward than those of
EEG. Patients only need to lie on the bed, and examiners attach ve
marker coils (some MEG systems require a lengthy procedure called
digitization of the head, but it is only optional for our system; RICOH
160-1). From a technical perspective, MEG avoids selection bias
associated with reference electrodes, a factor that always impacts EEG
data analysis. In addition, MEG is traditionally and theoretically
considered more accurate in the source estimations (Cohen etal.,
1990). However, the superiority of MEG is not empirically supported;
direct comparisons of the source estimation accuracies between MEG
and EEG (particularly high-density EEG) provided comparable results
(Cohen and Cun, 1991; Liu etal., 2002; Hedrich et al., 2017).
Furthermore, the oscillatory parameters are not dierent between
sensor- and source-levels. e source-level oscillatory parameters,
averaged within ve regions of interest (le frontal, right frontal, le
temporal, right temporal, and occipital), were shown to becorrelated
strongly with the sensor-level parameters, averaged within regions of
interest with spatial topographies similar to those used at the source-
level (Rodríguez-González etal., 2020). is suggests that oscillatory
parameters provide equivalent information at both sensor- and
source-levels, even when performing regional analyses. Consequently,
any potential advantage in the source estimation accuracy of MEG, if
present, may not provide additional benets in examining patients
with cognitive impairment using oscillatory parameters. Overall, both
FIGURE2
Within- and between-modality correlations among eZIS parameters,
NPAs, and age. Regression lines are added for significant
correlations. The scatterplot is colored dierently for each diagnosis
(black: healthy ageing, blue: MCI, red: AD, and grey: other types of
dementia). The number displayed at the corner of each plot indicates
Spearman’s coecient (rho) averaged across bootstrap iterations,
with an asterisk (*) indicating significant correlation. eZIS, easy
Z-score imaging system; NPAs, neuropsychological assessments;
MMSE, Mini-Mental State Examination; FAB, Frontal Assessment
Battery.
FIGURE3
Regional correlations between MEG oscillatory and eZIS parameters. Cortical projection of significant clusters (p < 0.025), indicated by colored regions
with correlation coecients between (A) RPt and severity and (B) IAF and severity. rho, Spearman’s coecient; MEG, magnetoencephalography; RPt,
relative power in theta band; IAF, individual alpha frequency; eZIS, easy Z-score imaging system.
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 09 frontiersin.org
MEG and EEG have comparable performance in the clinical practice
of electrophysiological examinations of patients with cognitive
impairments, which are currently selected based on the preferences of
researchers or practical reasons in each hospital/clinic. In this study,
our focus was on MEG, not only due to our preference for its
convenience and patient-friendliness, but also because recent
advancements in MEG hardware (e.g. optically pumped
magnetometers) are overcoming cost limitations and expanding
accessibility (Brookes etal., 2022). is development encourages the
active integration of MEG examinations into clinical practices in the
near future.
4.1 Associations between MEG oscillatory
and eZIS parameters
MEG has been used for the clinical examination of cognitive
impairment (Hoshi etal., 2022; Hirata et al., 2024), as it provides
oscillatory parameters sensitive to cognitive impairment. However,
its association with pathological conditions remains controversial.
According to the amyloid hypothesis of AD (Jack etal., 2010; Fagan,
2014; Bjorkli et al., 2020), neuronal/synaptic changes begin
following the appearance of amyloid-β (Aβ)-related biomarkers
(i.e. upstream biomarkers), such as Aβ levels detected through
cerebrospinal uid (CSF) measurements and positron emission
tomography (PET). ese changes precede structural alterations
(i.e. downstream biomarkers) measured using magnetic resonance
imaging (MRI) and computed tomography (CT). ese pathological
changes are followed by symptom manifestations (e.g. memory
impairment). erefore, changes in MEG oscillatory parameters
potentially reect pathological changes captured by midstream
biomarkers. Previous studies have demonstrated the associations
between MEG oscillatory characteristics and pathological
biomarkers, such as Aβ deposition, cortical tau burden, glucose
metabolism, synaptic density, and brain mass reduction (Fernández
etal., 2003; Nakamura etal., 2018; Coomans etal., 2021). Although
FIGURE4
Regional correlations between MEG oscillatory parameters and MMSE. Cortical projection of significant clusters (p < 0.025), indicated by colored
regions with correlation coecients between MMSE and (A) RPt, (B) RPa1, (C) RPa3, (D) RPb, (E) RPlg, (F) MF, (G) IAF, and (H) SSE. rho, Spearman’s
coecient; MEG, magnetoencephalography; RPt, relative power in theta band; RPa1, relative power in alpha1 band; RPa3, relative power in alpha3
band; RPb, relative power in beta band; RPlg, relative power in low gamma band; MF, mean frequency; IAF, individual alpha frequency; SSE, Shannon’s
spectral entropy; MMSE, Mini-Mental State Examination.
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 10 frontiersin.org
these studies have shown that MEG oscillatory parameters could
beused to identify pathological changes, they were conducted in
laboratory settings. It is important to investigate their relationship
with other neuroimaging biomarkers sensitive to pathological
conditions, currently used in clinical practice, to enhance the
validity and comprehensibility of MEG oscillatory parameters as
clinical examination tools for cognitive impairment.
is study examined these relationships using eZIS parameters as
clinically validated pathological biomarkers. e eZIS parameters are
neuroimaging biomarkers computed from SPECT measurements and
are commonly used in clinical practice to assess individuals with
cognitive decline (Matsuda etal., 2007; Imabayashi etal., 2017). e
eZIS parameters consist of three values, namely severity, extent, and
ratio, reecting the probability of AD. is study revealed the
following two main ndings: (a) RPt positively correlated with
severity, predominantly in the right hemisphere (Figures1, 3A), and
(b) IAF negatively correlated with severity in the caudal region
(Figure 3B), indicating that the slower shi of the alpha peak
corresponds to an increased likelihood of AD pathology.
e positive correlation between RPt and the eZIS parameter
related to AD (severity) was found mainly in the right hemisphere
(Figure3A), which was replicated using the dataset from individuals
with AD continuum (MCI + AD) (Supplementary Figure S4A;
Supplementary Table S4A). e severity provides quantication of
hypoperfusion in specic VOIs around the bilateral PCG, PC, and
parietal lobe, the enhancement of which is linked to an increased
pathological probability of AD (Matsuda etal., 2007). Its positive
correlation with RPt suggests that augmentation of theta power
corresponds to hypoperfusion in the VOIs, indicating a high
probability of AD. is nding corroborates the observation of a
previous EEG study of patients with AD, which demonstrated a close
relationship between rCBF and quantitative EEG parameters in the
theta band (Passero et al., 1995). Furthermore, a previous study
demonstrated a negative correlation between RP in the 2–6 Hz range
(spanning from delta to theta band) and global rCBF among both
patients with AD and individuals with healthy ageing, particularly in
the right hemisphere (Rodriguez etal., 1998). In a separate study, it
was found that RP in the 2.0–5.5 Hz range (delta to theta band)
FIGURE5
Regional correlations between MEG oscillatory parameters and FAB. Cortical projection of significant clusters (p < 0.025), indicated by colored regions
with correlation coecients between FAB and (A) RPt, (B) RPa1, (C) RPa3, (D) RPb, (E) RPlg, (F) RPhg, (G) MF, (H) IAF, and (I) SSE. rho, Spearman’s
coecient; MEG, magnetoencephalography; RPt, relative power in theta band; band; RPa1, relative power in alpha1 band; RPa3, relative power in
alpha3 band; RPb, relative power in beta band; RPlg, relative power in low gamma band; RPhg, relative power in high gamma band; MF, mean
frequency; IAF, individual alpha frequency; SSE, Shannon’s spectral entropy; FAB, Frontal Assessment Battery.
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 11 frontiersin.org
exhibited a negative correlation with rCBF in the parietal and right
hippocampal regions in patients with probable AD (Rodriguez etal.,
1999). Additionally, a more recent study on patients with MCI
demonstrated that a group with a low risk of conversion to AD showed
a negative correlation between RPt and rCBF in the hippocampal
complex (Moretti etal., 2013). Notably, enhancement of low-frequency
oscillatory activity, including the RPt, is a known MEG oscillatory
characteristic in patients with cognitive decline (Poza et al., 2007;
Fernández etal., 2013; López etal., 2014). Consistently, the present
study also demonstrated that global RPt negatively correlated with
NPAs (Figures 1, 4, 5), suggesting that RPt is sensitive to both
pathological and symptomatic changes accompanied by cognitive
impairment. e pathological background of RPt has been studied
using CSF biomarkers. For example, a previous study showed that RPt
in the resting-state EEG data was correlated with both NPAs and total
tau (T-tau) levels in CSF (Musaeus etal., 2018), while another EEG
study demonstrated a correlation between CSF Aβ-42 concentration
and the current source density over the right temporal area in the
theta band (Hata etal., 2016). Moreover, EEG studies revealed that the
combined ratio of the phosphorylated tau and Aβ42 (Aβ-42/p-tau
ratio) positively correlated with theta power in the right posterior
electrodes (Stomrud etal., 2010; Kramberger etal., 2013). A recent
EEG study also reported increased resting-state delta and theta
rhythms among patients with MCI with positive CSF biomarkers of
Aβ-42/p-tau ratio (Jovicich etal., 2019). e relationship between
cortical tau burden and the slowing of oscillatory activity in the
occipital region was also found in a combined study of MEG and [
18
F]
ortaucipir PET (Coomans etal., 2021). High CSF T-tau levels have
also been demonstrated to becorrelated with decreased rCBF in the
right superior posterior medial frontal lobe (Stomrud etal., 2012), and
it has been shown that the presence of Aβ deposition inuences
longitudinal changes in rCBF in humans (Sojkova etal., 2008) and
animal models (Maier etal., 2014). Additionally, the eZIS parameter
is sensitive to Aβ deposition (Takemaru etal., 2017). ese ndings
suggest that the increase in RPt in the right hemisphere reects the
presence of CSF biomarkers, including Aβ deposition and T-tau levels.
Furthermore, frontal and occipital theta power correlates with
hippocampal atrophy (Grunwald etal., 2007; Nakamura etal., 2018),
and global theta power correlates with memory function (Klimesch,
1999). erefore, RPt may not only serve as the sole reector of
pathological changes but also reect the symptomatic conditions of
cognitive impairment. Notably, the right hemisphere plays an
important role in the relationship between CSF biomarkers,
hypoperfusion, and oscillatory changes in the RPt (Stomrud etal.,
2010, 2012; Kramberger etal., 2013; Hata etal., 2016). Wepreviously
demonstrated that blood velocity was correlated with MF and IAF in
the right common carotid artery (CCA) but not in the le CCA
(Matsumoto et al., 2021). is nding suggests a hemispheric
asymmetry in the neuropathological mechanisms underlying AD.
Our second nding [i.e. negative correlations between IAF and
severity (Figure 3B)] indicates that changes in the alpha peak,
characterised by slow shis, are associated with hypoperfusion in the
VOIs and increased pathological probability of AD. is result was
also conrmed using the dataset from individuals with AD continuum
(MCI + AD) (Supplementary Figure S4B; Supplementary Table S4B).
is nding is consistent with that of a previous EEG study on patients
with AD, which showed a signicant positive correlation between the
rCBF in the temporoparietal region of interest and the peak
frequencies in the temporal and central electrodes (Kwa etal., 1993).
Another study on patients with mild to moderate AD demonstrated
that the MF, computed with a frequency range of 2–32 Hz, was
positively correlated with rCBF in the parietal lobule, including the
PC, superior parietal lobule, and postcentral gyrus (Rodriguez etal.,
2004). e denition of the MF in the previous study (i.e. a frequency
range of 2–32 Hz) (Rodriguez etal., 2004) was similar to that of the
IAF in this study (i.e. a frequency range of 4–15 Hz), which accounted
for the eects of a slow shi in the dominant frequency (i.e. alpha
peak). is similarity indicates that the ndings of the abovementioned
study are suggestive of a potential association between slow shi of the
alpha peak and hypoperfusion in the parietal lobule, including the PC,
superior parietal lobule, and postcentral gyrus (Rodriguez et al.,
2004). Notably, the relevance of alpha band power to the changes in
cholinergic neurotransmission, one of the AD pathologies, has been
reported previously. For example, the volumes of cholinergic cell
clusters corresponding to the medial septum, vertical and horizontal
limbs of the diagonal band, and posterior nucleus basalis of Meynert
positively correlated with pre-alpha (i.e. a frequency range of 5.5–8hz)
power in patients with MCI (Rea etal., 2021). Furthermore, an animal
study showed that alpha power was decreased by experimental
damage to this cholinergic pathway (Holschneider etal., 1998). A
1-year follow-up study demonstrated that a group of patients with
mild AD who responded to pharmacological treatment with donepezil
(i.e. acetylcholinesterase inhibitors) had a lesser magnitude reduction
of occipital and temporal alpha sources than their non-responder
counterparts (Babiloni et al., 2006b). e association between
cholinergic decits and EEG data was also examined using
scopolamine, a non-selective muscarine receptor antagonist that
blocks the stimulation of post-synaptic receptors. Aer scopolamine
administration, decreases in EEG alpha power were observed in
patients with AD and healthy controls (for reviews, see Ebert and
Kirch, 1998; Jeong, 2004). Similarly, another study demonstrated a
decrease in alpha power in the occipitotemporal regions in patients
with AD (Vecchio etal., 2011), which could bean electrophysiological
ngerprint of AD-specic pathological changes in cholinergic
neurotransmission.
4.2 Symptomatic relevance of MEG
oscillatory characteristics and eZIS
parameters
is study demonstrated that MEG oscillatory parameters
eectively capture the pathological changes represented by eZIS
parameters (Figures1, 3). Importantly, wefound that MEG oscillatory
parameters were not merely reectors of the pathological conditions
of patients. To explore their symptomatic signicance, weevaluated
the symptomatic relevance of MEG oscillatory and eZIS parameters
with NPAs. We demonstrated that MEG oscillatory parameters
correlated with both NPAs and eZIS parameters. However, direct
correlations between the NPAs and eZIS parameters were sparse
(Figure2).
This study showed that global slowing and loss of diversity in
neural oscillatory components were correlated with MMSE and
FAB scores. Although differences in MEG oscillatory parameters
between healthy ageing individuals and patients have been widely
studied (Engels etal., 2017; López-Sanz etal., 2018; Mandal etal.,
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 12 frontiersin.org
2018; Maestú etal., 2019), data on their links to NPAs are limited.
A previous study investigated the correlations between RPs and
various NPAs comprehensively (López et al., 2014), which
reported that MMSE was correlated negatively with RPd and RPt
and positively with RP in alpha band (8–12 Hz) and RPb.
Additionally, it revealed that performance of the trail making test,
an NPA for evaluating the executive functions, was correlated
negatively with RPd and RPt but positively with RP in alpha band
(8–12 Hz) and RPb (López et al., 2014). These results largely
corroborated our findings, with the exception of the alpha band.
Regarding the alpha band, wesubdivided the range into alpha1
(7–9 Hz), alpha2 (9–11 Hz), and alpha3 (11–13 Hz), following the
approach in a previous study (Wu and Liu, 1995). This
subdivision was motivated by the expectation that their
correlational behaviours with the level of cognitive impairments
would beopposite before and after the alpha peak (Poza etal.,
2007; Gómez etal., 2013; López etal., 2014); considering that the
bands as a whole might lead to cancelation of these opposing
effects. As expected, the behaviours of the lower alpha (alpha1)
and higher alpha (alpha3) were distinctive; the former
showed negative correlations, while the latter showed positive
correlations with the NPAs (Figures 1, 4B,C, 5B,C;
Supplementary Tables S7, S10B,C, S13B,C). In contrast, the
middle alpha (alpha2) around the peak of the alpha activity
showed no correlations with any of the parameters. The previous
study (López etal., 2014) used a whole alpha band (8–12 Hz) in
the correlational analysis but used subdivided alpha bands in the
group level comparison, indicating that a-md-MCI group, who
showed impairments in various cognitive domains, showed
smaller occipital RPa2 and right frontotemporal RP in 10–12 Hz
than for a-sd-MCI group, who exhibited an isolated memory
impairment. This difference implied that the cognitive functions,
except those in the memory domain, were responsive to the RPa2
and RP at 10–12 Hz. These results were also largely supported by
our results, which showed that RPa2 demonstrated positive
correlation coefficients with NPAs (Figure1), although they were
not statistically significant, and RPa3, which corresponds to
the RP in 10–12 Hz in the previous study, exhibited
positive correlations to the NPAs (Figures 1, 4C, 5C;
Supplementary Tables S7, S10C, S13C). Other MEG studies have
also reported a positive correlation between global MF and
MMSE scores (Fernández etal., 2006) and a negative correlation
between delta current densities and cognitive status (Fernández
etal., 2013). Additionally, another MEG study parameterised the
overall slowing of PSDn by calculating the ratio between RPs in
faster and slower bands (e.g. RP in the alpha band/RP in the theta
band) and showed that the slowing parameters were correlated
with the MMSE score as well as with the individual RPs (i.e. RPd,
RPt, and RPb) (Poza etal., 2008a). Notably, the ‘ratio’ parameters
of the oscillatory powers should beinterpreted with caution,
particularly when the study did not clarify the cause of the
changes, which may beattributed to the increase/decrease of
paired oscillatory powers, but there are alternative accounts; for
example, the ratio is heavily biased by the presence of an
aperiodic component (see section 4.3). A recent study by our
group showed that global MEG spectral parameters positively
correlated with MMSE/FAB scores (Hoshi et al., 2022). The
relationship between oscillatory parameters measured using EEG
and NPAs has been described previously. For example, previous
studies have shown negative correlations between parieto-
occipital delta sources and MMSE scores (Babiloni etal., 2006a;
Lizio etal., 2016), positive correlations between RPs in the alpha
and beta bands in frontal electrodes and MMSE scores
(Torabinikjeh etal., 2022), and positive correlations between the
prefrontal MF, IAF, alpha-to-theta ratio, and MMSE scores after
adjusting for age and education level (Choi et al., 2019).
Moreover, other studies examined the relationships between EEG
oscillatory parameters and MMSE scores in patients with
probable AD using a coefficient of determination (R
2
), which
quantifies the amount of data variation explained by MMSE;
these studies revealed that the changes in RPs in the theta, alpha,
and beta bands and SSE corresponded to changes in MMSE
scores (Garn et al., 2014, 2015; Coronel et al., 2017). These
findings, in line with our findings, indicate that the slowing and
loss of complexity/diversity in neural oscillatory signals are
associated with cognitive symptoms.
We found no significant correlations between the eZIS
parameters and NPAs. It is important to note that the eZIS
parameters—severity, extent, and ratio—were originally designed
to maximise discriminating performance between aMCI and
healthya ageing (i.e. maximising the area under the receiver
operator curve), where the NPAs were used as one of the selection
criteria of patients but not as parameters in the SPECT data
analysis. For maximising the discriminating performance, the
VOIs for the three parameters were set at the PCG, PC, and
parietal lobe (Matsuda etal., 2007). Previous studies have shown
that the MMSE score correlates with rCBF in the left hippocampus
(Ikeda etal., 2008); frontal, parietal, and medial temporal cortices
(Ushijima et al., 2002); and parietal and temporal cortices
(Kimura etal., 2012). Similarly, another study revealed that the
FAB score was correlated with rCBF in the left callosomarginal
and precentral regions (Yoshida etal., 2009). However, it is worth
noting that the VOIs used for computing the eZIS parameters (i.e.
the PCG and PC) did not overlap with the regions where previous
studies found correlations between rCBFs and MMSE or FAB
scores. This finding indicates that the eZIS parameters exclusively
capture hypoperfusion accompanied by pathological changes but
do not reflect functional changes associated with symptoms of
cognitive impairment. This notion is supported by a previous
study that demonstrated how treatment with donepezil, a
cholinesterase inhibitor used to effectively treat AD by enhancing
cholinergic neurotransmission (Birks and Harvey, 2018), led to
changes in rCBF in the PCC, along with changes in MMSE and
AD assessment scale-cognitive scale scores (Iizuka and
Kameyama, 2017). The eZIS parameters capture hypoperfusion
in PCC, reflecting pathological changes in AD responsive to
changes in cholinergic neurotransmission. The independence of
eZIS parameters from symptomatic measurements was also
exemplified by another study (Tokumitsu etal., 2021). The study
revealed that the discrimination performance of patients with
MCI and early AD was improved by subjecting MMSE and eZIS
parameter (extent) in the logistic regression model, compared to
the simple model including MMSE alone, indicating that the eZIS
parameter contained non-overlapping and additive information
to MMSE. The pathological changes do not always represent
dementia symptoms (Snowdon, 1997; Shiroky etal., 2007). The
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 13 frontiersin.org
three eZIS parameters may not becorrelated with NPAs as they
represent excerpts of rich information in rCBF, which were
designed to capture the pathological changes.
4.3 Limitations
This study has four limitations. First, the raw rCBF data (i.e.
whole brain data) were not used in the analysis for two reasons:
(1) this study focused on clinically validated pathological
biomarkers (i.e. eZIS parameters), and (2) the raw rCBF data
were not available in the clinical records of the hospital.
Therefore, the present results, such as the sparse correlations
between eZIS parameters and NPAs, cannot begeneralised to
rCBF data. The hypoperfusion severity in regions other than the
VOIs used for eZIS might capture symptomatic changes, but this
aspect remains to be addressed. Second, for examining
correlational relationships, the number of datasets was limited
(N= 64). However, wedid not calculate optimal sample size for
two reasons: (1) this was a retrospective study, and the number
of samples could not be modified following the power
calculations, and (2) estimating sample sizes requires expected
correlation coefficient sizes, which wecould not speculate on due
to the absence of previous studies examining the relationships
between MEG and eZIS parameters. We anticipate future
prospective studies with larger sample sizes. Third, this study did
not account for the presence of an aperiodic component in the
PSDn. The neural power spectra can be decomposed into a
periodic component, reflecting true oscillatory activity, and an
aperiodic component, which is 1/f-like activity modeled using a
Lorentzian function (Donoghue etal., 2020). Similar to classical
studies, this study assumed that changes in the neural power
spectra reflected the changes in oscillatory activities associated
with cognitive impairment. However, recent studies have
demonstrated the contribution of aperiodic component as a
predictor of cognitive impairment, which found that the
aperiodic components differed between AD vs. FTD (Wang etal.,
2023) and DLB/PD vs. MCI/control (Rosenblum etal., 2023),
while other studies showed no differences in the aperiodic
components among healthy controls, patients with MCI, and
patients with AD (Azami etal., 2023; Kopčanová etal., 2023). In
examining the correlations between RPs and NPAs, this study
demonstrated that the correlational directions were flipped
between low- (from RPd to RPa1) and high-frequency bands
(from RPa3 to RPhg), with a border transition at RPa2 (Figures1,
4, 5). This implies the contributions of the aperiodic component,
which shifts rotationally with a fulcrum around the peak
frequency (i.e. alpha band). Detailed analysis of our dataset,
which includes comparing aperiodic and periodic components,
will contribute to the unified understanding of the contradictory
findings in the previous studies. Notably, even if the correlations
were accounted by aperiodic components without contributions
of oscillatory components, it would not undermine the clinical
utility of the MEG oscillatory parameters. This is because the
presence of aperiodic contaminations would not change the
relationships between MEG oscillatory parameters and the
symptomatic/pathological information. The discriminations
between aperiodic and periodic components would beinfluential
only when we discuss the neural mechanisms behind the
correlations and interpretation of the results. Fourth, the MEG
oscillatory parameters were used ‘as is’ without refinement or
modification to examine their hindered relationships with the
eZIS parameters and NPAs. Fine-tuned parameters may reveal
associations not observed with the current settings. Wedid not
revise the MEG oscillatory parameters because they have been
used repeatedly in previous studies by us (Haraguchi etal., 2021;
Matsumoto etal., 2021; Hoshi etal., 2022; Hirata et al., 2024) and
others (Fernández etal., 2006; Poza etal., 2007, 2008a,b; Gómez
etal., 2013), and weaimed to examine THEIR relationships to
eZIS parameters, but not those of any other/new parameters. For
example, the MEG oscillatory parameters can be adjusted
individually; the International Federation of Clinical
Neurophysiology (IFCN)–EEG research workgroup (Babiloni
et al., 2020) encouraged adjusting the frequency bands
individually for minimising the biases on the statistical tests.
However, this requires modification of the existing MEG
oscillatory parameters, which is not addressed in this study. The
individual adjustment of the frequency bands would becrucial if
the study was addressing the pathological mechanisms underlying
the oscillatory changes in electrophysiological signals. The
parameters should be modified accordingly if future studies
address this research topic.
4.4 Conclusion
is study demonstrated that MEG oscillatory parameters
correlated with both eZIS parameters and NPAs. ese associations
between MEG oscillatory and eZIS parameters (clinically validated
pathological biomarkers) support the clinical validity of MEG
oscillatory parameters. Our results suggest that theta power
augmentation and slower shi of the alpha peak could serve as
potential ngerprints of AD pathology, and the global slowing and
loss of diversity in neural oscillatory components represent
symptomatic changes. Moreover, eZIS parameters did not show
correlations with NPAs. ese ndings demonstrate the potential of
MEG data to enhance the clinical examination of patients with
cognitive impairments, providing deeper insights into their clinical
status. e electrophysiological (MEG) examination uniquely shows
the symptomatic status of patients (i.e. NPAs), while pathological
(eZIS) examination provides clinically validated information about
the underlying cause of the cognitive impairments (i.e. AD). eir
combined use in the clinical practice would improve the overall care
of patients with cognitive impairments.
Data availability statement
e datasets presented in this study can befound in online
repositories. e names of the repository/repositories and accession
number(s) can befound at: Hoshi, Hideyuki (2023), ‘MEG x SPECT
study,’ Mendeley Data, V1, doi: 10.17632/h3bhhtxhjc.1.
Ethics statement
e studies involving humans were approved by Ethics Committee
of Kumagaya General Hospital. e studies were conducted in
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 14 frontiersin.org
accordance with the local legislation and institutional requirements.
Written informed consent for participation in this study was provided
by the participants’ legal guardians/next of kin.
Author contributions
HH: Conceptualisation, Data curation, Formal analysis, Investigation,
Methodology, Soware, Visualisation, Writing – original dra, Writing
– review & editing. YH: Investigation, Supervision, Writing – review &
editing. KF: Data curation, Investigation, Writing – review & editing. MK:
Investigation, Writing – review & editing. YS: Conceptualisation, Data
curation, Funding acquisition, Investigation, Project administration,
Supervision, Writing – original dra, Writing – review & editing.
Funding
e author(s) declare nancial support was received for the
research, authorship, and/or publication of this article. is study was
partially sponsored by RICOH Co., Ltd. e funder was not involved
in the study design, collection, analysis, interpretation of data, the
writing of this article or the decision to submit it for publication.
Acknowledgments
We thank Editage (www.editage.com) for providing English
language editing services. We also thank Mr. Hirokazu Shimizu, Mr.
Hideaki Kishibe (FUJIFILM Corporation), and Ms. Remiko Fujimoto
(PD Radiopharma Inc.) for their technical advice regarding the
SPECT system and its measurement and analysis protocols. We thank
Dr. Jesús Poza, Dr. Carlos Gómez, and Mr. Víctor Rodríguez-
González, and for their technical support with the
magnetoencephalography (MEG) analysis. Lastly, we thank our
patients and arm our genuine respect for their contributions to
continued progress in the medical sciences.
Conflict of interest
YS led the joint research projects supported by RICOH Co., Ltd.
HH was employed by RICOH Co., Ltd.
e remaining authors declare that the research was conducted in
the absence of any commercial or nancial relationships that could
beconstrued as a potential conict of interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their affiliated
organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or
claim that may be made by its manufacturer, is not guaranteed or
endorsed by the publisher.
Supplementary material
e Supplementary material for this article can befound online
at: https://www.frontiersin.org/articles/10.3389/fnagi.2024.1273738/
full#supplementary-material
References
Albert, M. S., DeKosky, S. T., Dickson, D., Dubois, B., Feldman, H. H., Fox, N. C., et al.
(2011). e diagnosis of mild cognitive impairment due to Alzheimer’s disease:
recommendations from the National Institute on Aging-Alzheimer’s association
workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 7,
270–279. doi: 10.1016/j.jalz.2011.03.008
American Psychiatric Association (2013). Diagnostic and statistical manual of mental
disorders. 5 Arlington, VA: e American Psychiatric Association.
Azami, H., Zrenner, C., Brooks, H., Zomorrodi, R., Blumberger, D. M., Fis cher, C. E.,
et al. (2023). Beta to theta power ratio in EEG periodic components as a potential
biomarker in mild cognitive impairment and Alzheimer’s dementia. Alzheimers Res.
e r. 15, 1–12. doi: 10.1186/S13195-023-01280-Z/FIGURES/5
Babiloni, C., Barry, R. J., Başar, E., Blinowska, K. J., Cichocki, A.,
Drinkenburg, W. H. I. M., et al. (2020). International Federation of Clinical
Neurophysiology (IFCN) – EEG research workgroup: recommendations on
frequency and topographic analysis of resting state EEG rhythms. Part 1:
applications in clinical research studies. Clin. Neurophysiol. 131, 285–307. doi:
10.1016/J.CLINPH.2019.06.234
Babiloni, C., Binetti, G., Cassarino, A., Dal Forno, G., Del Percio, C., Ferreri, F., et al.
(2006a). Sources of cortical rhythms in adults during physiological aging: a multicentric
EEG study. Hum. Brain Mapp. 27, 162–172. doi: 10.1002/hbm.20175
Babiloni, C., Cassetta, E., Dal Forno, G., Del Percio, C., Ferreri, F., Ferri, R., et al.
(2006b). Donepezil eects on sources of cortical rhythms in mild Alzheimer’s disease:
responders vs. non-responders. NeuroImage 31, 1650–1665. doi: 10.1016/J.
NEUROIMAGE.2006.02.015
Birks, J., and Harvey, R. (2018). Donepezil for dementia due to Alzheimer’s disease
(Review). Cochrane Database Syst. Rev. 2108:CD001190. doi: 10.1002/14651858.
CD001190.pub3
Bjorkli, C., Sandvig, A., and Sandvig, I. (2020). Bridging the gap between uid
biomarkers for Alzheimer’s disease, model systems, and patients. Front. Ag ing Neurosci.
12:568023. doi: 10.3389/FNAGI.2020.00272/BIBTEX
Blackman, R. B., and Tukey, J. W. (1958). e measurement of power spectra from the
point of view of communications engineering — part I. Bell Syst. Tech. J. 37, 185–282.
doi: 10.1002/j.1538-7305.1958.tb03874.x
Brookes, M. J., Leggett, J., Rea, M., Hill, R. M., Holmes, N., Boto, E., et al. (2022).
Magnetoencephalography with optically pumped magnetometers (OPM-MEG): the next
generation of functional neuroimaging. Trends Neurosci. 45, 621–634. doi: 10.1016/j.
tins.2022.05.008
Cai, C., Kang, H., Kirsch, H. E., Mizuiri, D., Chen, J., Bhutada, A., et al. (2019). Comparison
of DSSP and tSSS algorithms for removing artifacts from vagus nerve stimulators in
magnetoencephalography data. J. Neural Eng. 16:066045. doi: 10.1088/1741-2552/ab4065
Calamia, M., Markon, K., and Tranel, D. (2012). Scoring higher the second time
around: Meta-analyses of practice eects in neuropsychological assessment. Clin.
Neuropsychol. 26, 543–570. doi: 10.1080/13854046.2012.680913
Choi, J., Ku, B., You, Y. G., Jo, M., Kwon, M., Choi, Y., et al. (2019). Resting-state
prefrontal EEG biomarkers in correlation with MMSE scores in elderly individuals. Sci.
Rep. 9:10468. doi: 10.1038/s41598-019-46789-2
Cohen, D., and Cun, B. N. (1991). EEG versus MEG localization accuracy: theory
and experiment. Brain Topogr. 4, 95–103. doi: 10.1007/BF01132766/METRICS
Cohen, D., Cun, B. N., Yunokuchi, K., Maniewski, R., Purcell, C., Cosgrove, G. R.,
et al. (1990). MEG versus EEG localization test using implanted sources in the human
brain. Ann. Neurol. 28, 811–817. doi: 10.1002/ANA.410280613
Coomans, E. M., Schoonhoven, D. N., Tuncel, H., Verfaillie, S. C. J., Wolters, E. E.,
Boellaard, R., et al. (2021). In vivo tau pathology is associated with synaptic loss and
altered synaptic function. Alzheimers Res. er. 13:35. doi: 10.1186/S13195-021-00772-0
Coronel, C., Garn, H., Waser, M., Deistler, M., Benke, T., Dal-Bianco, P., et al. (2017).
Quantitative EEG markers of entropy and auto mutual information in relation to MMSE
scores of probable Alzheimer’s disease patients. Entropy 19:130. doi: 10.3390/E19030130
Davis, K. M., Ryan, J. L., Aaron, V. D., and Sims, J. B. (2020). PET and SPECT imaging
of the brain: history, technical considerations, applications, and radiotracers. Semin.
Ultrasound CT MRI 41, 521–529. doi: 10.1053/J.SULT.2020.08.006
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 15 frontiersin.org
de Winter, J. C. F., Gosling, S. D., and Potter, J. (2016). Comparing the Pearson and
Spearman correlation coecients across distributions and sample sizes: a tutorial using
simulations and empirical data. Psychol. Methods 21, 273–290. doi: 10.1037/MET0000079
Diamond, A. (2013). Executive functions. Annu. Rev. Psychol. 64, 135–168. doi:
10.1146/annurev-psych-113011-143750
Dinomais, M., Celle, S., Duval, G. T., Roche, F., Henni, S., Bartha, R., et al. (2016).
Anatomic correlation of the mini-mental state examination: a voxel-based
morphometric study in older adults. PLoS One 11:e0162889. doi: 10.1371/journal.
pone.0162889
Donoghue, T., Haller, M., Peterson, E. J., Varma, P., Sebastian, P., Gao, R., et al. (2020).
Parameterizing neural power spectra into periodic and aperiodic components. Nat.
Neurosci. 23, 1655–1665. doi: 10.1038/s41593-020-00744-x
Doppelmayr, M., Klimesch, W., Stadler, W., Pöllhuber, D., and Heine, C. (2002). EEG alpha
power and intelligence. Intelligence 30, 289–302. doi: 10.1016/S0160-2896(01)00101-5
Dubois, B., Slachevsky, A., Litvan, I., and Pillon, B. (2000). e FAB: a frontal
assessment battery at bedside. Neurology 55, 1621–1626. doi: 10.1212/WNL.55.11.1621
Ebert, U., and Kirch, W. (1998). Scopolamine model of dementia:
electroencephalogram ndings and cognitive performance. Eur. J. Clin. Investig. 28,
944–949. doi: 10.1046/J.1365-2362.1998.00393.X
Elahi, F. M., and Miller, B. L. (2017). A clinicopathological approach to the diagnosis
of dementia. Nat. Rev. Neurol. 13, 457–476. doi: 10.1038/NRNEUROL.2017.96
Engels, M. M. A., van der Flier, W. M., Stam, C. J., Hillebrand, A., Scheltens, P., and van
Straaten, E. C. W. (2017). Alzheimer’s disease: the state of the art in resting-state
magnetoencephalography. Clin. Neurophysiol. 128, 1426–1437. doi: 10.1016/j.
clinph.2017.05.012
Fagan, A. M. (2014). CSF biomarkers of Alzheimer’s disease: impact on disease
concept, diagnosis, and clinical trial design. Adv. Geriatr. 2014, 1–14. doi:
10.1155/2014/302712
Fernández, A., Arrazola, J., Maestú, F., Arno, C., Gil-Gregorio, P., Wienbruch, C., et al.
(2003). Correlations of hippocampal atrophy and focal low-frequency magnetic activity
in Alzheimer disease: volumetric MR imaging- magnetoencephalographic study. AJNR
Am. J. Neuroradiol. 24, 481–487.
Fernández, A., Hornero, R., Mayo, A., Poza, J., Gil-Gregorio, P., and Ortiz, T. (2006).
MEG spectral prole in Alzheimer’s disease and mild cognitive impairment. Clin.
Neurophysiol. 117, 306–314. doi: 10.1016/j.clinph.2005.10.017
Fernández, A., Tur rero, A., Zuluaga, P., Gil-Gregorio, P., Del Pozo, F., Maestu, F., et al.
(2013). MEG delta mapping along the healthy aging-alzheimer’s disease continuum:
diagnostic implications. J. Alzheimers Dis. 35, 495–507. doi: 10.3233/JAD-121912
Folstein, M. F., Folstein, S. E., and McHugh, P. R. (1975). “Mini-mental state”. A
practical method for grading the cognitive state of patients for the clinician. J. Psychiatr.
Res. 12, 189–198. doi: 10.1016/0022-3956(75)90026-6
Fonov, V., Evans, A., McKinstry, R., Almli, C., and Collins, D. (2009). Unbiased
nonlinear average age-appropriate brain templates from birth to adulthood. NeuroImage
47:S102. doi: 10.1016/s1053-8119(09)70884-5
Galasko, D., Abramson, I., Corey-Bloom, J., and al, L. J. (1993). Repe ated exposure
to the Mini-mental state examination and the information-memory-concentration test
results in a practice eect in Alzheimer’s diseas e. Neurology 43, 1559–1563. doi: 10.1212/
WNL.43.8.1559
Garn, H., Waser, M., Deistler, M., Benke, T., Dal-Bianco, P., Ransmayr, G., et al. (2015).
Quantitative EEG markers relate to Alzheimer’s disease severity in the prospective
dementia registry Austria (PRODEM). Clin. Neurophysiol. 126, 505–513. doi: 10.1016/J.
CLINPH.2014.07.005
Garn, H., Waser, M., Deistler, M., Schmidt, R., Dal-Bianco, P., Ransmayr, G., et al. (2014).
Quantitative EEG in Alzheimer’s disease: cognitive state, resting state and association with
disease severity. Int. J. Psychophysiol. 93, 390–397. doi: 10.1016/J.IJPSYCHO.2014.06.003
Gilbert, S. J., and Burgess, P. W. (2008). Exec utive function. Curr. Biol. 18, R110–R114.
doi: 10.1016/j.cub.2007.12.014
Giustiniani, A., Danesin, L., Bozzetto, B., Macina, A., Benavides-Varela, S., and
Burgio, F. (2023). Functional changes in brain oscillations in dementia: a review. Rev.
Neurosci. 34, 25–47. doi: 10.1515/REVNEURO-2022-0010/XML
Gómez, C., M Pérez-Macías, J., Poza, J., Fernández, A., and Hornero, R. (2013).
Spectral changes in spontaneous MEG activity across the lifespan. J. Neural Eng.
10:066006. doi: 10.1088/1741-2560/10/6/066006
Gramfort, A., Papadopoulo, T., Olivi, E., and Clerc, M. (2010). OpenMEEG:
Opensource soware for quasistatic bioelectromagnetics. Biomed. Eng. Online 9, 1–20.
doi: 10.1186/1475-925X-9-45
Grunwald, M., Hensel, A., Wolf, H., Weiss, T., and Gertz, H. J. (2007). Does the
hippocampal atrophy correlate with the cortical theta power in elderly subjects with a
range of cognitive impairment? J. Clin. Neurophysiol. 24, 22–26. doi: 10.1097/
WNP.0b013e31802ed5b2
Haraguchi, R., Hoshi, H., Ichikawa, S., Hanyu, M., Nakamura, K., Fukasawa, K., et al.
(2021). e menstrual cycle alters resting-state cortical activity: a
magnetoencephalography study. Front. Hum. Neurosci. 15:652789. doi: 10.3389/
fnhum.2021.652789
Hari, R., Baillet, S., Barnes, G., Burgess, R., Forss, N., Gross, J., et al. (2018). IFCN-
endorsed practical guidelines for clinical magnetoencephalography (MEG). Clin.
Neurophysiol. 129, 1720–1747. doi: 10.1016/j.clinph.2018.03.042
Hari, R., and Puce, A. (2017). MEG–EEG Primer. New York: Oxford Academic.
Hata, M., Tanaka, T., Kazui, H., Ishii, R., Canuet, L., Pascual-Marqui, R. D., et al. (2016).
Cerebrospinal uid biomarkers of Alzheimer’s disease correlate with electroencephalography
parameters assessed by exact low-resolution electromagnetic tomography (eLORETA). Clin.
EEG Neurosci. 48, 338–347. doi: 10.1177/1550059416662119
Hayashi, H., Kobayashi, R., Kawakatsu, S., Morioka, D., and Otani, K. (2020). Utility
of easy Z-score imaging system-assisted SPECT in detecting onset age-dependent
decreases in cerebral blood ow in the posterior cingulate cortex, precuneus, and
parietal lobe in Alzheimer’s disease with amyloid accumulation. Dement. Geriatr. Cogn.
Dis. Extra 10, 63–68. doi: 10.1159/000507654
Hedrich, T., Pellegrino, G., Kobayashi, E., Lina, J. M., and Grova, C. (2017).
Comparison of the spatial resolution of source imaging techniques in high-density EEG
and MEG. NeuroImage 157, 531–544. doi: 10.1016/j.neuroimage.2017.06.022
Heilbronner, R. L. (2007). American Academy of clinical neuropsychology (AACN)
practice guidelines for neuropsychological assessment and consultation. Clin.
Neuropsychol. 21, 209–231. doi: 10.1080/13825580601025932
Hirata, Y., Hoshi, H., Kobayashi, M., Shibamiya, K., Fukasawa, K., Ichikawa, S.,
et al (2024). Monitoring the outcomes of non‐pharmacological treatments for cognitive
impairment using magnetoencephalography: A case series. Clin. Case Rep. 12, e8385.
doi: 10.1002/ccr3.8385
Holschneider, D. P., Leuchter, A. F., Scremin, O. U., Treiman, D. M., and Walton, N. Y.
(1998). Eects of cholinergic deaerentation and NGF on brain electrical coherence.
Brain Res. Bull. 45, 531–541. doi: 10.1016/S0361-9230(97)00446-2
Hoshi, H., Hirata, Y., Kobayashi, M., Sakamoto, Y., Fukasawa, K., Ichikawa, S., et al.
(2022). Distinctive eects of executive dysfunction and loss of learning/memory abilities
on resting-state brain activity. Sci. Rep. 12:3459. doi: 10.1038/s41598-022-07202-7
Iadecola, C. (2017). e neurovascular unit coming of age: a journey through
neurovascular coupling in health and disease. Neuron 96, 17–42. doi: 10.1016/J.
NEURON.2017.07.030
Iizuka, T., and Kameyama, M. (2017). Cholinergic enhancement increases regional
cerebral blood ow to the posterior cingulate cortex in mild Alzheimer’s disease. Geriatr
Gerontol Int 17, 951–958. doi: 10.1111/GGI.12818
Ikeda, E., Shiozaki, K., Takahashi, N., Togo, T., Odawara, T., Oka, T., et al. (2008). Total
Mini-mental state examination score and regional cerebral blood ow using Z score
imaging and automated ROI analysis soware in subjects with memory impairment.
Ann. Nucl. Med. 22, 539–542. doi: 10.1007/s12149-007-0148-2
Imabayashi, E., Soma, T., Sone, D., Tsukamoto, T., Kimura, Y., Sato, N., et al. (2017).
Validation of the cingulate island sign with optimised ratios for discriminating dementia
with Lewy bodies from Alzheimer’s disease using brain perfusion SPECT. Ann. Nucl.
Med. 31, 536–543. doi: 10.1007/s12149-017-1181-4
Imabayashi, E., Yokoyama, K., Tsukamoto, T., Sone, D., Sumida, K., Kimura, Y., et al.
(2016). e cingulate island sign within early Alzheimer’s disease-specic hypoperfusion
volumes of interest is useful for dierentiating Alzheimer’s disease from dementia with
Lewy bodies. EJNMMI Res. 6, 67–68. doi: 10.1186/s13550-016-0224-5
Jack, C. R., Knopman, D. S., Jagust, W. J., Shaw, L. M., Aisen, P. S., Weiner, M. W., et al.
(2010). Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological
cascade. Lancet Neurol. 9, 119–128. doi: 10.1016/S1474-4422(09)70299-6
Jagust, W. J., Johnson, K. A., and Holman, B. L. (1995). SPECT perfusion imaging in
the diagnosis of dementia. J. Neuroimaging 5, s45–s52. doi: 10.1111/JON19955S1S45
Jeong, J. (2004). EEG dynamics in patients with Alzheimer’s disease. Clin.
Neurophysiol. 115, 1490–1505. doi: 10.1016/j.clinph.2004.01.001
Jovicich, J., Babiloni, C., Fer rari, C., Marizzoni, M., Moretti, D. V., Del Percio, C., et al.
(2019). Two-year longitudinal monitoring of amnestic mild cognitive impairment
patients with prodromal Alzheimer’s disease using topographical biomarkers derived
from functional magnetic resonance imaging and electroencephalographic activity. J.
Alzheimers Dis. 69, 15–35. doi: 10.3233/JAD-180158
Kale, R. U., Ingale, P. M., Murade, R. T., and Sayyad, S. S. (2013). Comparison of
quality power spectrum estimation (Bartlett, Welch, Blackman & Tukey) methods. Int.
J. Sci. Mod. Eng. 1, 28–31.
Kimura, N., Kumamoto, T., Masuda, T., Hanaoka, T., Okazaki, T., and Arakawa, R.
(2012). Evaluation of the regional cerebral blood ow changes during long-term
donepezil therapy in patients with Alzheimer’s disease using 3DSRT. J. Neuroimaging
22, 299–304. doi: 10.1111/J.1552-6569.2011.00612.X
Klimesch, W. (1999). EEG alpha and theta oscillations reect cognitive and memory
performance: a review and analysis. Brain Res. Rev. 29, 169–195. doi: 10.1016/
S0165-0173(98)00056-3
Kopčanová, M., Tait, L., Donoghue, T., Stothart, G., Smith, L., Sandoval, A. A. F., et al.
(2023). Resting-state EEG signatures of Alzheimer’s disease are driven by periodic but
not aperiodic changes. bioRxiv 2023.06.11.544491. doi: 10.1101/2023.06.11.544491
Kramberger, M. G., Kåreholt, I., Andersson, T., Winblad, B., Eriksdotter, M., and
Jelic, V. (2013). Association between EEG abnormalities and CSF biomarkers in a
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 16 frontiersin.org
memory clinic cohort. Dement. Geriatr. Cogn. Disord. 36, 319–328. doi:
10.1159/000351677
Kwa, V. I. H., Weinstein, H. C., Posthumus Meyjes, E. F., Van Royen, E. A., Bour, L. J.,
Verhoe, P. N. L. G., et al. (1993). Spectral analysis of the EEG and 99m-Tc-HMPAO
SPECT-scan in Alzheimer’s disease. Biol. Psychiatry 33, 100–107. doi:
10.1016/0006-3223(93)90308-Z
Kybic, J., Clerc, M., Faugeras, O., Keriven, R., and Papadopoulo, T. (2005). Fast
multipole acceleration of the MEG/EEG boundary element method. Phys. Med. Biol. 50,
4695–4710. doi: 10.1088/0031-9155/50/19/018
Larson-Prior, L. J., Oostenveld, R., Della Penna, S., Michalareas, G., Prior, F.,
Babajani-Feremi, A., et al. (2013). Adding dynamics to the human connectome project
with MEG. NeuroImage 80, 190–201. doi: 10.1016/J.NEUROIMAGE.2013.05.056
Lin, F. H., Witzel, T., Ahlfors, S. P., Stuebeam, S. M., Belliveau, J. W., and
Hämäläinen, M. S. (2006). Assessing and improving the spatial accuracy in MEG source
localisation by depth-weighted minimum-norm estimates. NeuroImage 31, 160–171.
doi: 10.1016/J.NEUROIMAGE.2005.11.054
Liu, A. K., Dale, A. M., and Belliveau, J. W. (2002). Monte Carlo simulation studies of
EEG and MEG localization accuracy. Hum. Brain Mapp. 16, 47–62. doi: 10.1002/
HBM.10024
Lizio, R., Del Percio, C., Marzano, N., Soricelli, A., Yener, G. G., Basąr, E., et al. (2016).
Neurophysiological assessment of Alzheimer’s disease individuals by a single
electroencephalographic marker. J. Alzheimers Dis. 49, 159–177. doi: 10.3233/
JAD-143042
López, M. E., Cuesta, P., Garcés, P., Castellanos, P. N., Aurtenetxe, S., Bajo, R., et al.
(2014). MEG spectral analysis in subtypes of mild cognitive impairment. Age (Omaha)
36, 9624–1112. doi: 10.1007/s11357-014-9624-5
López-Sanz, D., Serrano, N., and Maestú, F. (2018). e role of
magnetoencephalography in the early stages of Alzheimer’s disease. Front. Neurosci.
12:572. doi: 10.3389/fnins.2018.00572
Maestú, F., Cuesta, P., Hasan, O., Fernandéz, A., Funke, M., and Schulz, P. E. (2019).
e importance of the validation of M/EEG with current biomarkers in Alzheimer’s
disease. Front. Hum. Neurosci. 13:17. doi: 10.3389/fnhum.2019.00017
Maier, F. C., Wehrl, H. F., Schmid, A. M., Mannheim, J. G., Wiehr, S., Lerdkrai, C.,
et al. (2014). Longitudinal PET-MRI reveals β-amyloid deposition and rCBF dynamics
and connects vascular amyloidosis to quantitative loss of perfusion. Nat. Med. 20,
1485–1492. doi: 10.1038/nm.3734
Makeig, S., Jung, T., Ghahremani, D., and Sejnowski, T. (1996). Independent component
analysis of simulated ERP data. Tech. Rep. INC-9606. Institute for Neural Computation,
San Diego, CA.
Malek, N., Baker, M. R., Mann, C., and Greene, J. (2017). Electroencephalographic
markers in dementia. Acta Neurol. Scand. 135, 388–393. doi: 10.1111/ANE.12638
Mandal, P. K., Banerjee, A., Tripathi, M., and Sharma, A. (2018). A comprehensive
review of magnetoencephalography (MEG) studies for brain functionality in healthy
aging and Alzheimer’s disease (AD). Front. Comput. Neurosci. 12:60. doi: 10.3389/
fncom.2018.00060
Maris, E., and Oostenveld, R. (2007). Nonparametric statistical testing of EEG- and
MEG-data. J. Neurosci. Methods 164, 177–190. doi: 10.1016/j.jneumeth.2007.03.024
Martínez-Cagigal, V. (2021). Multiple testing toolbox MATLAB Central File Exchange
Available at: https://www.mathworks.com/matlabcentral/leexchange/70604-multiple-
testing-toolbox.
Matsuda, H., Mizumura, S., Nagao, T., Ota, T., Iizuka, T., Nemoto, K., et al. (2007).
Automated discrimination between very early Alzheimer disease and controls using an
easy Z-score imaging system for multicenter brain perfusion single-photon emission
tomography. AJNR Am. J. Neuroradiol. 28, 731–736.
Matsumoto, T., Hoshi, H., Hirata, Y., Ichikawa, S., Fukasawa, K., Gonda, T., et al. (2021).
e association between carotid blood ow and resting-state brain activity in patients with
cerebrovascular diseases. Sci. Rep. 11:15225. doi: 10.1038/s41598-021-94717-0
McCarey, R. J., and Westervelt, H. J. (1995). Issues associated with repeated
neuropsychological assessments. Neuropsychol. Rev. 5, 203–221. doi: 10.1007/
BF02214762
McKeith, I. G., Boeve, B. F., Dickson, D. W., Halliday, G., Taylor, J. P., Weintraub, D.,
et al. (2017). Diagnosis and management of dementia with Lewy bodies. Neurology 89,
88–100. doi: 10.1212/WNL.0000000000004058
Mizumura, S., and Kumita, S. I. (2006). Stereotactic statistical imaging analysis of the
brain using the easy Z-score imaging system for sharing a normal database. Radiat. Med.
24, 545–552. doi: 10.1007/s11604-006-0056-8
Moretti, D. V., Prestia, A., Binetti, G., Zanetti, O., and Frisoni, G. B. (2013). Increase
of theta frequency is associated with reduction in regional cerebral blood ow only in
subjects with mild cognitive impairment with higher upper alpha/low alpha EEG
frequency power ratio. Front. Behav. Neurosci. 7:188. doi: 10.3389/FNBEH.2013.00188/
BIBTEX
Musaeus, C. S., Engedal, K., Høgh, P., Jelic, V., Mørup, M., Naik, M., et al. (2018). EEG
theta power is an early marker of cognitive decline in dementia due to Alzheimer’s
disease. J. Alzheimers Dis. 64, 1359–1371. doi: 10.3233/JAD-180300
Nakamura, A., Cuesta, P., Fernández, A., Arahata, Y., Iwata, K., Kuratsubo, I., et al.
(2018). Electromagnetic signatures of the preclinical and prodromal stages of
Alzheimer’s disease. Brain 141, 1470–1485. doi: 10.1093/BRAIN/AWY044
Oostenveld, R., Fries, P., Maris, E., and Schoelen, J.-M. (2011). FieldTrip: open source
soware for advanced analysis of MEG, EEG, and invasive electrophysiological data.
Comput. Intell. Neurosci. 2011:156869. doi: 10.1155/2011/156869
Passero, S., Rocchi, R., Vatti, G., Burgalassi, L., and Baitistini, N. (1995). Quantitative
EEG mapping, regional cerebral blood ow, and neuropsychological function in
Alzheimer’s disease. Dementia 6, 148–156. doi: 10.1159/000106938
Poza, J., Hornero, R., Abásolo, D., Fernández, A., and García, M. (2007). Extraction
of spectral based measures from MEG background oscillations in Alzheimer’s disease.
Med. Eng. Phys. 29, 1073–1083. doi: 10.1016/J.MEDENGPHY.2006.11.006
Poza, J., Hornero, R., Abásolo, D., Fernández, A., and Mayo, A. (2008a). Evaluation of
spectral ratio measures from spontaneous MEG recordings in pat ients with Alzheimer’s
disease. Comput. Methods Prog. Biomed. 90, 137–147. doi: 10.1016/j.cmpb.2007.12.004
Poza, J., Hornero, R., Escudero, J., Fernández, A., and Sánchez, C. I. (2008b). Regional
analysis of spontaneous MEG rhythms in patients with alzheimer’s dis ease using spectral
entropies. Ann. Biomed. Eng. 36, 141–152. doi: 10.1007/s10439-007-9402-y
Rea, R. C., Berlot, R., Martin, S. L., Craig, C. E., Holmes, P. S., Wright, D. J., et al.
(2021). Quantitative EEG and cholinergic basal forebrain atrophy in Parkinson's
disease and mild cognitive impairment. Neurobiol. Aging 106, 37–44. doi: 10.1016/J.
NEUROBIOLAGING.2021.05.023
Rizkallah, J., Amoud, H., Fraschini, M., Wendling, F., and Hassan, M. (2020).
Exploring the correlation between M/EEG source-space and fMRI networks at rest.
Brain Topogr. 33, 151–160. doi: 10.1007/S10548-020-00753-W
Rodriguez, G., Nobili, F., C opello, F., Vitali, P., Gianelli, M. V., Taddei, G., et al. (1999).
99mTc-HMPAO regional cerebral blood ow and quantitative electroencephalography
in Alzheimer’s disease: a correlative study. J. Nucl. Med. 40, 522–529.
Rodriguez, G., Nobili, F., Rocca, G., De Carli, F., Gianelli, M. V., and Rosadini, G.
(1998). Quantitative electroencephalography and regional cerebral blood ow:
discriminant analysis between Alzheimer’s patients and healthy controls. Dement.
Geriatr. Cogn. Disord. 9, 274–283. doi: 10.1159/000017072
Rodriguez, G., Vitali, P., Canfora, M., Calvini, P., Girtler, N., De Leo, C., et al. (2004).
Quantitative EEG and perfusional single photon emission computed tomography
correlation during long-term donepezil therapy in Alzheimer’s disease. Clin.
Neurophysiol. 115, 39–49. doi: 10.1016/S1388-2457(03)00321-3
Rodríguez-González, V., Gómez, C., Shigihara, Y., Hoshi, H., Revilla-Vallejo, M.,
Hornero, R., et al. (2020). Consistency of local activation parameters at sensor- and
source-level in neural signals. J. Neural Eng. 17:056020. doi: 10.1088/1741-2552/abb582
Rolls, E. T., Huang, C. C., Lin, C. P., Feng, J., and Joliot, M. (2020). Automated anatomical
labelling atlas 3. NeuroImage 206:116189. doi: 10.1016/J.NEUROIMAGE.2019.116189
Rosenblum, Y., Shiner, T., Bregman, N., Giladi, N., Maidan, I., Fahoum, F., et al.
(2023). Decreased aperiodic neural activity in Parkinson’s disease and dementia with
Lewy bodies. J. Neurol. 270, 3958–3969. doi: 10.1007/S00415-023-11728-9/FIGURES/2
Sachdev, P. S., Blacker, D., Blazer, D. G., Ganguli, M., Jeste, D. V., Paulsen, J. S., et al.
(2014). Classifying neurocognitive disorders: the DSM-5 approach. Nat. Rev. Neurol. 10,
634–642. doi: 10.1038/nrneurol.2014.181
Sekihara, K., Kawabata, Y., Ushio, S., Sumiya, S., Kawabata, S., Adachi, Y., et al. (2016). Dual
signal subspace projection (DSSP): a novel algorithm for removing large interference in
biomagnetic measurements. J. Neural Eng. 13:036007. doi: 10.1088/1741-2560/13/3/036007
Shigemori, K., Ohgi, S., Okuyama, E., Shimura, T., and Schneider, E. (2010). e
factorial structure of the mini mental state examination (MMSE) in Japanese dementia
patients. BMC Geriatr. 10, 1–7. doi: 10.1186/1471-2318-10-36
Shiroky, J. S., Schipper, H. M., Bergman, H., and Chertkow, H. (2007). Can youhave
dementia with an MMSE score of 30? Am. J. Alzheimers Dis. Other Dement. 22, 406–415.
doi: 10.1177/1533317507304744
Snowdon, D. A. (1997). Aging and Alzheimer’s disease: lessons from the Nun study.
Gerontologist 37, 150–156. doi: 10.1093/GERONT/37.2.150
Sojkova, J., Beason-Held, L., Zhou, Y., An, Y., Kraut, M. A., Ye, W., et al. (2008).
Longitudinal cerebral blood ow and amyloid deposition: An emerging pattern? J. Nucl.
Med. 49, 1465–1471. doi: 10.2967/JNUMED.108.051946
Stomrud, E., Forsberg, A., Hägerström, D., Ryding, E., Blennow, K., Zetterberg, H.,
et al. (2012). CSF biomarkers correlate with cerebral blood ow on SPECT in healthy
elderly. Dement. Geriatr. Cogn. Disord. 33, 156–163. doi: 10.1159/000338185
Stomrud, E., Hansson, O., Minthon, L., Blennow, K., Rosén, I., and Londos, E. (2010).
Slowing of EEG correlates with CSF biomarkers and reduced cognitive speed in elderly
with normal cognition over 4 years. Neurobiol. Aging 31, 215–223. doi: 10.1016/j.
neurobiolaging.2008.03.025
Tadel, F., Baillet, S., Mosher, J. C., Pantazis, D., and Leahy, R. M. (2011). Brainstorm:
a user-friendly application for MEG/EEG analysis. Comput. Intell. Neurosci. 2011:879716.
doi: 10.1155/2011/879716
Tait, L., Özkan, A., Szul, M. J., and Zhang, J. (2021). A systematic evaluation of source
reconstruction of resting MEG of the human brain with a new high-resolution atlas:
Hoshi et al. 10.3389/fnagi.2024.1273738
Frontiers in Aging Neuroscience 17 frontiersin.org
performance, precision, and parcellation. Hum. Brain Mapp. 42, 4685–4707. doi:
10.1002/HBM.25578
Takemaru, M., Kimura, N., Abe, Y., Goto, M., and Matsubara, E. (2017). e
evaluation of brain perfusion SPECT using an easy Z-score imaging system in the mild
cognitive impairment subjects with brain amyloid-β deposition. Clin. Neurol. Neurosurg.
160, 111–115. doi: 10.1016/J.CLINEURO.2017.06.018
Tokumitsu, K., Yasui-Furukori, N., Takeuchi, J., Yachimori, K., Sugawara, N.,
Terayama, Y., et al. (2021). The combination of MMSE with VSRAD and eZIS has
greater accuracy for discriminating mild cognitive impairment from early
Alzheimer’s disease than MMSE alone. PLoS One 16:e0247427. doi: 10.1371/
JOURNAL.PONE.0247427
Torabinikjeh, M., Asayesh, V., Dehghani, M., Kouchakzadeh, A., Marhamati, H., and
Gharibzadeh, S. (2022). Correlations of frontal resting-state EEG markers with MMSE
scores in patients with Alzheimer’s disease. Egypt. J. Neurol. Psychiatry Neurosurg. 58,
1–7. doi: 10.1186/S41983-022-00465-X/TABLES/5
Ushijima, Y., Okuyama, C., Mori, S., Nakamura, T., Kubota, T., and Nishimura, T. (2002).
Relationship between cognitive function and regional cerebral blood ow in Alzheimer’s
disease. Nucl. Med. Commun. 23, 779–784. doi: 10.1097/00006231-200208000-00012
Vecchio, F., Babiloni, C., Lizio, R., De Vico Fallani, F., Blinowska, K., Verrienti, G.,
et al. (2013). Resting state cortical EEG rhythms in Alzheimer’s disease: toward EEG
markers for clinical applications: a review. Suppl. Clin. Neurophysiol. 62, 223–236. doi:
10.1016/B978-0-7020-5307-8.00015-6
Vecchio, F., Lizio, R., Frisoni, G. B., Ferri, R., Rodriguez, G., and Babiloni, C. (2011).
Electroencephalographic rhythms in alzheimer’s disease. Int. J. Alzheimers Dis.
2011:927573. doi: 10.4061/2011/927573
Wahlstrom, M., and Boersma, F. J. (1968). e inuence of test-wiseness upon
achievement. Educ. Psychol. Meas. 28, 413–420. doi: 10.1177/001316446802800221
Wang, Z., Hospital, Z., Liu, A., Yu, J., Wang, P., Bi, Y., et al. (2023). e eect of
aperiodic components in distinguishing Alzheimer’s disease from frontotemporal
dementia. Res. Sq. [Preprint]. doi: 10.21203/rs.3.rs-2915225/v1
Wu, X., and Liu, X. Q. (1995). Study of the alpha frequency band of healthy adults in
quantitative EEG. Clin. EEG Neurosci. 26, 131–136. doi: 10.1177/155005949502600212/
ASSET/155005949502600212.FP.PNG_V03
Yoshida, H., Terada, S., Sato, S., Kishimoto, Y., Ata, T., Ohshima, E., et al. (2009).
Frontal assessment battery and brain perfusion imaging in early dementia. Dement.
Geriatr. Cogn. Disord. 27, 133–138. doi: 10.1159/000198687
