ISSN 0006-2979, Biochemistry (Moscow), 2026, Vol. 91, No. 5, pp. 770-779 © The Author(s) 2026. This article is an open access publication.
770
Method for Individual Assessment of Human
Cerebral Cortex Activity by a Combined Use
of Magnetic Resonance Spectroscopy
of Glutamate and BOLD Signal Method
Aleksandr D. Korotkov
1,a
*, Artem D. Myznikov
1
, Ilya M. Krasnov
1
,
Mikhail D. Didur
1
, Denis V. Cherednichenko
1
, and Maxim V. Kireev
2
1
N. P. Bechtereva Institute of the Human Brain, Russian Academy of Sciences, 197022 St.Petersburg, Russia
2
Institute for Cognitive Studies, Saint Petersburg State University, 199004 St. Petersburg, Russia
a
e-mail: korotkov@ihb.spb.ru
Received January 21, 2026
Revised March 30, 2026
Accepted March 31, 2026
Abstract— Modern magnetic resonance imaging (MRI) methods enable individualized assessment of both
functional brain activity and neurochemical composition. Functional magnetic resonance imaging (fMRI)
allows evaluation of brain activity at rest and during task performance, while magnetic resonance spectros-
copy (MRS) provides measurements of key metabolites such as choline, N-acetylaspartate, creatine, lactate,
lipids, alanine, glutamine and glutamate, GABA, and myo-inositol. These approaches are widely used in both
fundamental brain research and diagnostic studies. However, existing literature lacks methods for directly
comparing these individual assessments, which is essential for investigating relationships between metabolite
levels and brain activity. Here, we present a method for aligning individual fMRI and MRS data. Using this
approach, we demonstrated a neurophysiological phenomenon in which the functional connectivity between
brain regions increases while overall functional activity decreases during task performance.
DOI: 10.1134/S0006297926600213
Keywords: glutamate, fMRI, BOLD signal, connectomics, functional connectivity, magnetic resonance spectroscopy
* To whom correspondence should be addressed.
INTRODUCTION
Methodological advances and accumulated ex-
perience in functional tomographic neuroimaging
of human brain, particularly functional magnetic
resonance imaging (fMRI), now make it possible, in
principle, to conduct individually oriented studies
relevant to both basic scientific research and clinical
diagnostics. In most scientific investigations, howev-
er, individual data are averaged at the group level,
whereas single-subject fMRI analysis is typically lim-
ited to diagnostic applications.
Much of the current understanding of human
brain functions comes from activation-based studies.
At the same time, it is well established that human
behavior is supported by dynamically organized, spa-
tially distributed functional neuroanatomical systems
[1-4], consisting of nodes and connections between
them. Incorporating this framework at the method-
ological level in modern functional connectomics
methods has demonstrated that task-related reorgani-
zation of these systems during task engagement gen-
erally involves either combined or isolated changes
in the local activity and functional connectivity: task
engagement is typically associated with increases,
and disengagement with decreases, in these param-
eters. There is evidence that when a component of a
neuroanatomical system becomes engaged in activity,
its functional connectivity may change even in the
absence of alterations in the local activity. This ob-
servation is consistent with the logic of the activa-
tion–connectivity framework. In contrast, it has been
reported that the functional connectivity between
the network nodes might increase upon reduction
GLUTAMATE AND BRAIN SYSTEMS 771
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
in the blood oxygen level-dependent (BOLD) signal.
This phenomenon has been observed across a broad
range of studies on visual information categorization
[5], speech activity [6], episodic memory [7], working
memory [8], social interaction [9], and other forms
of higher nervous activity. Such findings are difficult
to reconcile with the standard activation–connectiv-
ity framework and, given the still unresolved com-
plexities in interpreting the relationship between the
BOLD signal and the levels of metabolic activity in
human brain [10, 11], raise a question of whether
the observed decrease in the BOLD signal truly re-
flects a decrease in the functional activity of the cor-
responding brain system nodes. So far, this issue has
received little attention, despite its importance for
both personalized diagnostics and pathophysiological
research. In particular, reduced activity is often in-
terpreted as a marker of dysfunction (e.g., cognitive
decline), which emphasizes the need to clarify the
contribution of brain structure to supporting normal
task performance.
Specifically, we detected a decrease in the BOLD
signal accompanied by strengthening of the function-
al connectivity in the angular gyrus, one of the re-
gions of the “social” brain, during social interaction
tasks[12,13]. This raises an important question: does
a reduction in the BOLD signal in this context truly
reflect decreased functional activity?
One way to address this question is through di-
rect assessment of brain metabolic activity. Magnetic
resonance spectroscopy (MRS) provides this oppor-
tunity by enabling measurement of neurotransmit-
ter (e.g., glutamate) concentrations reflecting cortical
activation levels.
Glutamate, the major excitatory neurotransmitter
in the human central nervous system, plays a princi-
pal role in the formation of BOLD response that can
be measured by fMRI. Biochemical processes under-
lying the hemodynamic response development are
collectively known as neurovascular coupling and
depend on the coordinated interactions between as-
trocytes, pericytes, and smooth muscle cells [14]. Glu-
tamate released into the synaptic binds to its cognate
receptors on the postsynaptic membrane, triggering
an increase in intracellular calcium and activation of
specific signaling cascades, ultimately leading to the
release of vasodilatory agents such as nitric oxide
(NO), prostaglandin E2 (PgE2), and others. These sub-
stances, in turn, indirectly act on smooth muscle cells
through pericytes and astrocytes, causing vasodilation
and increased local cerebral blood flow [14]. Recent
meta-analysis and systematic review of functional
MRS studies reported consistent increase in glutamate
concentration across a range of task paradigms, in-
cluding motor and visual tasks [15]. However, rela-
tively few studies have examined glutamate dynamics
in situations when the BOLD signal decreased during
task engagement, and these studies mainly observed
reduction in the glutamate concentration during visu-
al stimulation [16].
The question posed can be addressed by using a
combination of fMRI with MRS for measuring gluta-
mate concentration. However, even when the anatom-
ical location of the structure or region of interest (ROI)
is known, the methodological differences between the
two MRI modalities complicate direct comparison of
the obtained results: MRS data are acquired within
a relatively large rectangular voxel whose size is of-
ten exceeds the size of the fMRI-identified activation
cluster. Together with substantial interindividual vari-
ability in the location of BOLD signal changes, this
requires individual alignment of fMRI and MRS data.
To our knowledge, no prior studies have described
an analytical approach for directly comparing indi-
vidual estimates of the BOLD signal magnitude and
glutamate concentration.
In light of these challenges, the aim of the pres-
ent study was to develop a method for aligning in-
dividual data on the BOLD signal level (assessed by
fMRI) and glutamate concentration (assessed by MRS).
Using this approach, we further sought to determine
whether local functional activity decreases in a brain
region that, during task performance, demonstrates
both reduced BOLD signal and increased functional
connectivity with other brain structures.
MATERIALS AND METHODS
Study participants. Two volunteers participated
in this study (women aged 19 and 20 years). Neither
participant reported a history of neurological or psy-
chiatric disease. Both were right-handed, as assessed
using the Edinburgh Handedness Inventory[17]. Both
volunteers provided written informed consent to par-
ticipate in the study. All procedures were approved
by the Ethics Committee of the Bechtereva Institute
of the Human Brain.
Data acquisition. The protocol for the neuroim-
aging data acquisition is illustrated in Fig. 1. Struc-
tural MRI was first performed to acquire high-res-
olution T1-weighted images, followed by MRS data
acquisition in the ROIs during wakeful rest, with
eyes open and fixation on a cross. Next, fMRI scan-
ning and MRS while participants performed the task
described below.
Stimuli and procedure. The study used a mod-
ified and expanded Russian-language adaptation of
the Reading the Mind in the Eyes Test (RMET). Partic-
ipants were presented with photographs of the facial
region around the eyes that were displayed at the
center of the screen, accompanied by four adjectives
KOROTKOV et al.772
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
Fig. 1. Study design (WI, weighted images).
placed one in each corner (for example, haughty, de-
pressed, confident, and satisfied). Their task was to
select the adjective that best described the facial ex-
pression and, accordingly, the affective or psycholog-
ical state of the depicted individual. The images were
taken from a stimulus set originally developed at the
McGill University [18]. The images were cropped and
processed to include only the region around the eyes,
consistent with the standard RMET format. For each
image, a set of four adjectives was fixed, with includ-
ed one correct answer and three fixed alternatives.
The correspondence and its correspondence between
the images and response options was validated in an
independent behavioral study[12]. In the control con-
dition, participants viewed the same photographs of
the eye region, but instead of mental state adjectives,
they were presented with four age options (e.g., twen-
ty, thirty, forty, fifty, and sixty) and were instructed
to estimate the age of the depicted person. In both
the experimental and control conditions, the positions
of the four response options were randomized across
trials to prevent systematic association between the
spatial location and response choice. All options were
selected from a common pool, with no repetition
within a single presentation.
The experimental task included 144 unique
stimuli. Each stimulus was presented twice: once in
the emotion recognition task and once in the age
judgment task. No stimuli were repeated within the
same condition, resulting in a total of 288 trials.
The test consisted of three sessions, each lasting
17 min. Before each session, a fixation cross (white
cross on a black background) was displayed for 5.5 s.
Each session included alternating task and rest blocks
with the fixation cross presentation (Fig. 2). Each ses-
sion contained 24 task blocks and 24 rest blocks, with
12 blocks for each condition (emotion recognition
and age judgment).
Before each task block, an instruction cue (“Eyes”
or “Age”) was presented for 2 s. The fixation crosses
then appeared on the screen for a variable time in-
terval ranging from 1.5 to 3.5 s, with an increment
of 0.5 s.
The task block consisted of four trials, each last-
ing 5 s. During each trial, participants were shown
a photograph of the eye region with four response
options (either adjectives or age alternatives) and
had to indicate their choice by pressing a joystick
button. The intertrial interval varied from 1 to 4 s,
with a step of 1 s, resulting in a total task block du-
ration of 30 s. Rest blocks were presented between
the task blocks, and their duration varied from 2
to 12 s, with an increment of 2 s. Stimulus presen-
tation, recording of the response and reaction time,
Fig. 2. Scheme of emotion recognition task (RMET).
GLUTAMATE AND BRAIN SYSTEMS 773
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
Fig. 3. Regions showing increased (yellow) and decreased (pale blue) BOLD signal (based on data from [12]) along with
strengthened functional connectivity between during RMET. Red and blue, anatomical regions of increased and decreased
BOLD signal, respectively, that also showed increased functional connectivity with the regions of decreased BOLD signal
during RMET. Statistical maps were constructed using an uncorrected threshold of p < 0.001 followed by the cluster-level
family-wise error (FWE) correction at p<0.05. Red arrows indicate pairs of brain regions with significantly increased func-
tional connectivity during RMET (based on data from [13]). ROIs for MRS analysis are indicated in blue: L_AG_A39(PGa)_TPJ
and R_AG_A39(PGa)_TPJ. A – anterior, P – posterior, S – superior, L – left.
and synchronization with the fMRI image acquisition
were implemented using an MR-compatible Nordic-
Neurolab system and E-Prime2.0 software (Psychology
Software Tools, USA).
MRS and fMRI data acquisition. All MRI data
were acquired using a General Electric SIGNA Archi-
tect 3.0T MRI scanner equipped with a 48-channel
head radiofrequency coil. To minimize head motion
during scanning, participants’ heads were stabilized
with an MR-compatible head and neck brace (Schanz
collar).
Structural images were acquired using a
T1-weighted pulse sequence [T1-MPRAGE; repetition
time (TR), 2605 ms; echo time (TE), 2.4 ms; inversion
time (TI), 1000 ms; flip angle, 8 degrees]. A total of
180 axial slices were obtained with a slice thickness
of 0.9 mm and a pixel size of 0.9×0.9 mm [field of
view (FOV), 240×240 mm; scan matrix, 256×256).
Functional T2* images were acquired during the
task performance using a single-shot echo-planar im-
aging (EPI) sequence (TE, 21.8 ms; flip angle, 52 de-
grees; FOV, 208×208 mm; scan matrix, 86×86). Con-
tinuously acquired images consisted of 48 axial slices
2.8 mm thick, with a voxel size of 2.4×2.4×2.8 mm,
and covered the entire cerebral cortex and most of
the cerebellum, with slice orientation aligned to the
structural images. The acquisition time for one func-
tional image (TR) was 960 ms. The acquisition time
for one task session was 16 min 45 s, corresponding
to 1046 dynamic scans. Resting-state fMRI data were
acquired for 6min 24 s, corresponding to 400 dynam-
ic scans. fMRI data were collected for the entire brain
volume.
MRS was performed using a standard single-
voxel PRESS (Point-RESolved Spectroscopy) pulse se-
quence with the following parameters: TE, 35 ms; TR,
1500 ms; 128 spectral averages; 16 phase cycles; wa-
ter suppression using CHESS (CHEmical Shift Selective
saturation); bandwidth, 5000 Hz; and 4096 points. The
cubic voxel volume was 8 mL (2×2×2 cm), which is
standard for single-voxel MRS studies [15]. Spectro-
scopic volumes were positioned in the angular gyri in
accordance with the aim of the study, namely target-
ing the regions identified as exhibiting reduced BOLD
signal alongside strengthened functional connectivity
with other brain areas (Fig. 3).
In addition, within the same pulse sequence, two
phase cycles were acquired with the registration of
unsuppressed water signal; these were subsequently
used for MRS data preprocessing and quantitative
evaluation. Spectroscopic data were collected both
during wakeful rest with fixation (prior to fMRI
during the RMET) and during the RMET (following
the fMRI described above) (Fig. 1).
KOROTKOV et al.774
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
Preprocessing and quantitative analysis of
MRS data. MRS data were preprocessed using the
Osprey 2.9.6 software package in MATLAB R2023b
(Mathworks, USA). The preprocessing steps included:
(i) phase alignment of individual spectroscopic exci-
tations in the time domain; (ii) averaging of individ-
ual spectroscopic excitations; (iii) Fourier transfor-
mation; (iv) removal of residual water signal in the
range 4.6-4.8 ppm; (v) linear baseline correction; and
(vi) eddy current correction. For each spectrum, qual-
ity parameters were obtained, including the signal-to-
noise ratio (SNR) and the full width at half maximum
(FWHM) of the creatine peak (spectra with the cre-
atine peak FWHM exceeding 12 Hz or creatine peak
SNR below 10 were excluded from further analysis).
The spectra were then fitted using the Osprey lin-
ear-combination model with a basis set of spectra for
individual metabolites. The mean relative amplitude
residual (MRAR) was used to assess the quality of
the fit. In parallel, spectroscopic volume masks were
generated and used for subsequent segmentation into
gray matter, white matter, and cerebrospinal fluid.
Data preprocessing and calculation of absolute
glutamate concentrations were performed in Osprey.
The glutamate peak at 2.35 ppm was selected for
quantification. The calculation took into account the
ratio between different brain tissue fractions (gray
matter, white matter, and cerebrospinal fluid) and
was carried out by segmenting the structural imag-
es within the spectroscopic volume and estimating
the relative content of each fraction. The concentra-
tions were then calculated with allowance for these
relative fractions based on the information obtained
from the MR spectra acquired with the unsuppressed
water signal according to the method described by
Gasparovic et al. [19]. The resulting glutamate con-
centrations (mM) were compared using the percent-
age change calculated according to the formula (1):
% = (C
task
− C
rest
) ÷ C
rest
× 100, (1)
where C
rest
and C
task
are glutamate concentrations(mM)
rest and during task performance, respectively.
Statistical analysis of fMRI data was performed
according to the procedure described in [12]. Due to
the discrepancy between the directions of group ef-
fects observed by the activation fMRI (decrease in the
local activity) and changes in the glutamate concen-
tration in the second participant (increase instead of
expected decrease; see Results), we performed addi-
tional analysis of the activation fMRI data. A general
linear model was constructed using a finite impulse
response (FIR) with a 30-second time window and
31 time points to evaluate the profile of the BOLD
response. As an integral measure task-related BOLD
signal changes independent of the test condition,
we used the total area under the curve (AUC) of the
BOLD signals extracted from each voxel of the spec-
troscopic volume according to the FIR model. At the
first stage, beta coefficients for all FIR regressors were
extracted from each voxel of fMRI images included
in the spectroscopic volume mask for both age judg-
ment and emotion recognition conditions. AUC values
were then calculated for the BOLD signals obtained
from each voxel using the trapz function in MATLAB,
which computes the integral via the trapezoidal rule
with unit spacing. At this stage, negative AUC values
(arbitrary units) indicated a predominance of a de-
crease in the BOLD signal, whereas positive AUC val-
ues corresponded to its increase. Finally, the AUC val-
ues for both test conditions were summed to provide
a single measure reflecting the overall predominance
of either decrease or increase in the BOLD signal
throughout the entire fMRI session. For visualization,
density histograms were constructed to display the
distribution of voxels across the AUC values.
Analysis of functional connectivity based on
fMRI data. To compare the obtained data on gluta-
mate concentration and BOLD signal level with the
functional connectivity measures during task perfor-
mance, we used the task-modulated functional con-
nectivity (TMFC) method [20, 21]. Changes in TMFC
were analyzed for the ROI corresponding to the spec-
troscopic volume using the TMFC software developed
by our group (https://github.com/IHB-IBR-department/
TMFC_toolbox). TMFC was calculated using the gen-
eralized psychophysiological interaction (gPPI) frame-
work [22]. To exclude the influence of coactivations
on the TMFC estimates, statistical models were im-
plemented using the FIR approach. Analogous to the
analysis of BOLD signal changes, individual GLMs
were constructed to identify significant differences
between the ROI (spectroscopic volume) and all other
voxels of the functional brain images in the strength
of functional connectivity between the emotion rec-
ognition and age judgment conditions (see [13] for a
detailed protocol of TMFC analysis). Statistical para-
metric maps were constructed for contrasting TMFC
statistical parameters (beta coefficients) of the “Emo-
tion Recognition > Age Judgment” type, with an un-
corrected threshold of p < 0.001 followed by the FWE
correction for multiple comparisons at the cluster
level (p < 0.05).
RESULTS
All spectra included in analysis met the crite-
ria described above (mean creatine peak FWHM,
5.26 ± 0.48 Hz; mean creatine peak SNR, 50.4 ± 5.79).
The MRAR following basis-set fitting was 1.72 ± 0.21%.
Glutamate concentrations were assessed before and
GLUTAMATE AND BRAIN SYSTEMS 775
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
Fig. 4. MRS results: glutamate concentration in the left and
right angular gyri (lAG and rAG, respectively).
during RMET performance (Fig. 4). The first partici-
pant showed expected decrease in the glutamate con-
centration in both right and left angular gyri (by 6.6%
and 2.2%, respectively) relative to the group effect.
However, the second participant exhibited increases
(although of different magnitudes) in both ROIs: by
6.9% in the right angular gyrus and by 0.6% in the
left angular gyrus.
The observed discrepancy between change in the
glutamate concentration and decrease in the activity
observed in the group analysis of fMRI data prompt-
ed additional analysis of individual activation fMRI
data following the approach detailed in the “Statis-
tical Analysis of fMRI Data” section. The total AUC
for the spectroscopic voxels of both participants and
the percentage changes in glutamate concentration
are presented in Fig. 5. For the first participant, re-
duced glutamate levels corresponded to a negative
total AUC, indicating a decrease in the BOLD signal
during the task session. Conversely, the second par-
ticipant showed increase in the glutamate concen-
tration alongside positive AUC values were obtained,
reflecting an increase in the BOLD signal, which is
fully consistent with current models describing the
relationship between glutamate dynamics and BOLD
signal.
The results of the functional connectivity analy-
sis for the ROI in which the above-described patterns
in the relationship between the BOLD signal level
and glutamate concentration were identified, are
shown in Fig. 5. With the exception of the ROI in the
left angular gyrus of participant 2, all ROIs exhibited
stronger functional connectivity during the emotion
recognition task compared to the age judgment task.
A consistent finding was strengthened connectivity
with the precuneus (indicated by crosshair in Fig. 6).
DISCUSSION
A combined use of fMRI and MRS in this study
enabled us to establish a link between the direction
of BOLD signal changes and alterations in glutamate
levels. even in the presence of substantial interindi-
vidual variability in the BOLD signal. Moreover, these
changes were accompanied by strengthening of the
Fig. 5. Density histograms of the AUCs for all fMRI voxels within the spectroscopic volume: orange, emotion recognition;
blue, age judgment; red, intersection for emotion recognition and age judgement; a.u., arbitrary units.
KOROTKOV et al.776
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
Fig. 6. TMFC increases during emotion recognition. Colored areas are clusters of statistically significant increases in TMFC
detected during emotion recognition compared with age judgment. Gray crosshair marks the brain region for which this
increase in TMFC was observed across all significant results of analysis.
long-range functional connections of the analyzed
ROIs, which also shared a common spatial location
in the precuneus. Whereas similar effects had previ-
ously been reported only in group studies, our results
demonstrate them at the level of individual analysis
and provide initial evidence that the greater involve-
ment of a network node in functional interactions
[12, 13], despite a decrease in the BOLD signal, is
associated with the signs of deactivation. Specifical-
ly, such regions exhibit decreased concentrations of
the excitatory neurotransmitter glutamate, suggesting
that the active involvement of a node in interactions,
despite a reduced BOLD signal, indeed reflects a low-
er level of excitation. For the first time, a coupling
between changes in the glutamate concentration and
BOLD signal has been demonstrated for a complex
and socially significant activity in a region of the as-
sociative cerebral cortex.
The results of the present study showed concor-
dance between the direction of changes in the BOLD
signal and glutamate concentration measured by MRS
in both analyzed cases by using a carefully designed
method to relate them in terms of spatial location
of BOLD signal changes and the spectroscopic voxel.
However, previous studies have reported discrepan-
cies between these measures, and several hypotheses
have been proposed to account for such inconsisten-
cies. Glutamate plays a central role in both synaptic
transmission and metabolic processes [23]. Through
amidation, glutamate is converted to glutamine,
which serves both as a transport form of ammonia
and a substrate that can enter the tricarboxylic acid
(TCA) cycle, forming the basis of the glutamate-gluta-
mine cycle[24]. Thus, the principal pathways of gluta-
mate metabolism include: (i) entry into the TCA cycle
via oxidative deamination of glutamate to alpha-ke-
toglutarate to meet the cell’s energy demands [23];
(ii) synthesis of glutamate from glutamine via hydro-
lytic deamination mediated by phosphate-activated
glutaminase and transport via the malate-aspartate
shuttle, enabling its role as a neurotransmitter [24];
(iii) use of glutamine as an ammonia reservoir with
subsequent inclusion in the ornithine cycle for de-
toxification [25]; and (iv) conversion of glutamic acid
into GABA synthesis via decarboxylation [26].
On one hand, these biochemical processes re-
flect functional differences in glutamate metabol-
ic pathways; on the other hand, they influence the
availability of glutamate for detection by MRS, since
each process proceeds in a specific cellular compart-
ment and, therefore, in a different microenvironment
[27-30]. Given a close relationship between glutamine
and glutamate, distinguishing between their levels in
MRS remains a key challenge, a technical solution for
which has yet to be established. Nevertheless, studies
using 7 T scanners demonstrate that such separation
GLUTAMATE AND BRAIN SYSTEMS 777
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
is feasible, enabling acquisition of a glutamate signal
without glutamine contamination [31].
Another set of factors that may play a role in
the relationship between the BOLD signal and MRS
data involves functional differences between gluta-
mate receptors. Two glutamate receptor families are
currently recognized: ionotropic and metabotropic.
The former includes N-methyl-D-aspartate (NMDA),
alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic
acid (AMPA), and kainate receptors; the latter include
three subclasses of G protein-coupled receptors [32].
Each glutamate receptor group exerts specific effects
on the CNS structures, which may substantially in-
fluence the observed BOLD signal. For example, in a
study by Gsell et al. [33], administration of an NMDA
receptor antagonist caused a delayed but irreversible
decrease in the BOLD response to stimulation. In con-
trast, administration of an AMPA receptor antagonist
produced an immediate and pronounced decrease in
the hemodynamic response, with recovery occurring
within 30 min after administration [33].
Analysis of the published data reveals substan-
tial inconsistencies across the studies examining the
BOLD signal and MRS data, particularly in terms of
acquisition protocols, analytical approaches, conclu-
sions, and interpretation[15]. These unresolved issues
substantiate the need for further methodological de-
velopment for better relating the BOLD signal to the
neurotransmitter concentration evaluated by MRS.
The data obtained in our study, within the proposed
framework for integrating individual indicators of lo-
cal activity and neurotransmitter concentration, offer
a potential way for addressing such inconsistencies.
The results indicate the promise of the developed
approach for individualized functional diagnostics,
which extends far beyond the task addressed in the
present study. MRS enables the quantification of mul-
tiple metabolites, including choline, N-acetylaspartate,
creatine, lactate, lipids, alanine, glutamine, glutamate,
GABA, and myo-inositol, whereas fMRI assesses the
functional activity in brain structures. Both methods
can be applied during resting state or during engage-
ment in various activities and are suitable for moni-
toring these indicators over time, for example during
treatment. Overall, our findings not only suggest ave-
nues for more detailed investigation of brain system
reorganization in independent group studies, but also
point to directions for methodological advancements
aimed at translating this approach into personalized
clinical diagnostics.
CONCLUSION
The method we developed for aligning MRS and
fMRI data allowed us to relate an individual meta-
bolic parameter – glutamate concentration – to the
functional brain activity, thereby addressing the
following research question. Across two individual
fMRI–MRS studies, glutamate concentrations were as-
sessed during the RMET in ROIs in the angular gyri
of the left and right hemispheres and compared with
individual BOLD signal values in these structures un-
der RMET conditions. This comparison showed that
the direction of BOLD signal changes corresponded to
changes in the glutamate concentration, even in the
presence of interindividual variability of fMRI signals
in the analyzed regions. These findings provide initial
evidence that the previously observed cases of great-
er involvement of a system node in the functional
interactions, despite reduced fMRI signal, are char-
acterized by the signs of deactivation. Specifically,
the levels of the excitatory neurotransmitter in such
structures are reduced.
Based on the obtained data, the proposed ap-
proach involving a transition from group functional
connectomic data to the identification of individual
brain regions for the analysis of neurotransmitter
concentration, holds promise for individualized diag-
nostics. Further research is needed to validate this
approach.
Abbreviations
AUC area under the curve
BOLD blood oxygen-level dependent
TMFC task-modulated functional connectivity
fMRI functional magnetic resonance imaging
FIR finite impulse response
FWE family-wise error
MRS magnetic resonance spectroscopy
RMET Reading the Mind in the Eyes Test
ROI region of interest
Contributions
A.D.K., M.V.K., and D.V.Ch. developed the concept and
supervised the study; A.D.M. and I.M.K. performed the
experiments and collected data; M.D.D., D.V.Ch., and
M.V.K. discussed the results; A.D.K., A.D.M., I.M.K., and
M.V.K., wrote the text of the article; M.D.D. and D.V.Ch.
edited the manuscript.
Funding
This study was supported by the Russian Science
Foundation (project no. 23-18-00521).
Ethics approval and consent to participate
All procedures involving human participants were in
accordance with the ethical standards of the Nation-
al Research Ethics Committee and the 1964 Helsinki
Declaration and its later amendments or comparable
ethical standards. Voluntary informed consent was
obtained from each participant included in the study.
KOROTKOV et al.778
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
Conflict of interest
The authors of this work declare that they have no
conflicts of interest.
Open access
This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits
use, sharing, adaptation, distribution, and reproduc-
tion in any medium or format, as long as you give
appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons li-
cense, and indicate if changes were made. The im-
ages or other third party material in this article are
included in the article’s Creative Commons license,
unless indicated otherwise in a credit line to the
material. If material is not included in the article’s
Creative Commons license and your intended use
is not permitted by statutory regulation or exceeds
the permitted use, you will need to obtain permis-
sion directly from the copyright holder. To view a
copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
REFERENCES
1. Fornito, A. (2023) Brain organization: from cells and circuits to systems and networks, in Handbook of Neuro-
psychology, Volume2: Neuroscience and neuromethods (Brown,G.G., Crosson, B., Haaland, K.Y., and King, T. Z.,
eds) American Psychological Association, Washington, p. 3-32, https://doi.org/10.1037/0000308-001.
2. Bekhterev, V. M. (1896) Conduction Pathways of the Spinal Cord and Brain: A Guide to the Study of the Internal
Connections of the Brain/[Works of] V. M. Bekhterev, Professor of the Military Medical Academy, 2nd Edn.,
thoroughly revised and substantially expanded. Parts 1-2, K. L. Rikker, St. Petersburg [in Russian].
3. Bekhtereva, N. P. (1971) Neurophysiological aspects of human mental activity, Meditsina, Leningrad.
4. Medvedev, S. V., and Pakhomov, S. V. (1989) Dynamic organization of brain systems, Nauka, Leningrad.
5. Medvedev, S. V., Masharipov, R. S., Korotkov, A. D., and Kireev, M. V. (2023) Characteristics of the involvement
of hidden nodes in the activity of human brain systems revealed on fMRI data, Hum. Physiol., 49, 1-11, https://
doi.org/10.1134/S0362119722700141.
6. Kireev, M., Slioussar, N., Korotkov, A. D., Chernigovskaya, T. V., and Medvedev, S. V. (2015) Changes in functional
connectivity within the fronto-temporal brain network induced by regular and irregular Russian verb produc-
tion, Front. Hum. Neurosci., 9, 36, https://doi.org/10.3389/fnhum.2015.00036.
7. Gerchen, M. F., and Kirsch, P. (2017) Combining task‐related activation and connectivity analysis of fMRI data re-
veals complex modulation of brain networks, Hum. Brain Mapp., 38, 5726-5739, https://doi.org/10.1002/hbm.23762.
8. Satake, T., Taki, A., Kasahara, K., Yoshimaru, D., and Tsurugizawa, T. (2024) Comparison of local activation,
functional connectivity, and structural connectivity in the N-back task, Front. Neurosci., 18, 1337976, https://
doi.org/10.3389/fnins.2024.1337976.
9. Zheltyakova, M., Korotkov, A., Masharipov, R., Myznikov, A., Didur, M., Cherednichenko, D., Wagels, L., Habel, U.,
Kireev, M., and Votinov, M. (2022) Social interaction with an anonymous opponent requires increased involve-
ment of the theory of mind neural system: an fMRI study, Front. Behav. Neurosci., 16, 807599, https://doi.org/
10.3389/fnbeh.2022.807599.
10. Epp, S., Castrillón, G., Yuan, B., Andrews-Hanna, J., Preibisch, C., and Riedl, V. (2023) Two distinct modes of
hemodynamic responses in the human brain, bioRxiv, https://doi.org/10.1101/2023.12.08.570806.
11. Mishra, A.M., Ellens, D.J., Schridde,U., Motelow, J. E., Purcaro, M.J., DeSalvo, M.N., Enev,M., Sanganahalli,B.G.,
Hyder,F., and Blumenfeld,H. (2011) Where fMRI and electrophysiology agree to disagree: corticothalamic and stria-
tal activity patterns in the WAG/Rij rat, J.Neurosci., 31, 15053-15064, https://doi.org/10.1523/JNEUROSCI.0101-11.2011.
12. Zheltyakova, M. A., Myznikov, A. D., Korotkov, A. D., Knyazeva, I. S., Masharipov, R. S., Cherednichenko, D. V.,
and Kireev, M. V. (2025) Brain activity during the recognition of emotions from eye expressions, Neurosci. Be-
hav. Physi., 55, 1131-1142, https://doi.org/10.1007/s11055-025-01870-4.
13. Zheltyakova,M., Kireev, M. V., Myznikov, A., Cherednichenko, D., Didur, M., and Korotkov, A. (2026) Brain Human
network for reading the mind from the eyes: beyond activated and deactivated regions, SCAN, in press.
14. Hall, C. N., Howarth, C., Kurth-Nelson, Z., and Mishra, A. (2016) Interpreting BOLD: towards a dialogue between
cognitive and cellular neuroscience, Philos. Trans. R. Soc. Lond. B Biol. Sci., 371, 20150348, https://doi.org/
10.1098/rstb.2015.0348.
15. Pasanta, D., He, J. L., Ford, T., Oeltzschner, G., Lythgoe, D. J., and Puts, N. A. (2023) Functional MRS studies of
GABA and glutamate/Glx – a systematic review and meta-analysis, Neurosci. Biobehav. Rev., 144, 104940, https://
doi.org/10.1016/j.neubiorev.2022.104940.
16. Boillat, Y., Xin, L., Van Der Zwaag, W., and Gruetter, R. (2020) Metabolite concentration changes associated with
positive and negative BOLD responses in the human visual cortex: a functional MRS study at 7 Tesla, J. Cereb.
Blood Flow Metab., 40, 488-500, https://doi.org/10.1177/0271678X19831022.
GLUTAMATE AND BRAIN SYSTEMS 779
BIOCHEMISTRY (Moscow) Vol. 91 No. 5 2026
17. Oldfield, R. C. (1971) The assessment and analysis of handedness: the Edinburgh inventory, Neuropsychologia,
9, 97-113, https://doi.org/10.1016/0028-3932(71)90067-4.
18. Schmidtmann, G., Jennings, B., Sandra, D., Pollock, J., and Gold, I. (2020) The McGill Face database: validation
and insights into the recognition of facial expressions of complex mental states, Perception, 49, 310-329, https://
doi.org/10.1177/0301006620901671.
19. Gasparovic, C., Song, T., Devier, D., Bockholt, H. J., Caprihan, A. G., Mullins, P. G., Posse, S., Jung, R. E., and
Morrison, L. A. (2006) Use of tissue water as a concentration reference for proton spectroscopic imaging, Magn.
Reson. Med., 55, 1219-1226, https://doi.org/10.1002/mrm.20901.
20. Masharipov, R., Knyazeva, I., Korotkov, A., Cherednichenko, D., and Kireev, M. (2024) Comparison of whole-brain
task-modulated functional connectivity methods for fMRI task connectomics, Commun Biol., 7, 1402, https://
doi.org/10.1038/s42003-024-07088-3.
21. Masharipov, R. S., Didour, M. D., Cherednichenko, D. V., and Kireev, M. V. (2026) Studying task-modulated func-
tional connectivity using functional magnetic resonance imaging, Neurosci. Behav. Physiol., 56, 98-114, https://
doi.org/10.1007/s11055-026-01954-9.
22. McLaren, D. G., Ries, M. L., Xu, G., and Johnson, S. C. (2012) A generalized form of context-dependent psy-
chophysiological interactions (gPPI): a comparison to standard approaches, Neuroimage, 61, 1277-1286, https://
doi.org/10.1016/j.neuroimage.2012.03.068.
23. Schousboe, A. (2017) A Tribute to Mary C. McKenna: glutamate as energy substrate and neurotransmitter –
functional interaction between neurons and astrocytes, Neurochem. Res., 42, 4-9, https://doi.org/10.1007/s11064-
015-1813-9.
24. Andersen, J. V., and Schousboe, A. (2023) Milestone review: metabolic dynamics of glutamate and GABA medi-
ated neurotransmission – the essential roles of astrocytes, J. Neurochem., 166, 109-137, https://doi.org/10.1111/
jnc.15811.
25. Cooper, A. J. L. (2001) Role of glutamine in cerebral nitrogen metabolism and ammonia neurotoxicity, Ment.
Retard. Dev. Disabil. Res. Rev., 7, 280-286, https://doi.org/10.1002/mrdd.1039.
26. Martin, D. L., and Rimvall, K. (1993) Regulation of γ‐aminobutyric acid synthesis in the brain, J. Neurochem.,
60, 395-407, https://doi.org/10.1111/j.1471-4159.1993.tb03165.x.
27. Li, H., Rodríguez-Nieto, G., Chalavi, S., Seer, C., Mikkelsen, M., Edden, R. A. E., and Swinnen, S. P. (2024) MRS-as-
sessed brain GABA modulation in response to task performance and learning, Behav. Brain Funct., 20, 22,
https://doi.org/10.1186/s12993-024-00248-9.
28. Craven, A. R., Dwyer, G., Ersland, L., Kazimierczak, K., Noeske, R., Sandøy, L. B., Johnsen, E., and Hugdahl, K.
(2024) GABA, glutamatergic dynamics and BOLD contrast assessed concurrently using functional MRS during a
cognitive task, NMR Biomed., 37, e5065, https://doi.org/10.1002/nbm.5065.
29. Ip, I. B., Emir, U. E., Parker, A. J., Campbell, J., and Bridge, H. (2019) Comparison of neurochemical and BOLD
signal contrast response functions in the human visual cortex, J. Neurosci., 39, 7968-7975, https://doi.org/10.1523/
JNEUROSCI.3021-18.2019.
30. Ip, I. B., Berrington, A., Hess, A. T., Parker, A. J., Emir, U. E., and Bridge, H. (2017) Combined fMRI-MRS ac-
quires simultaneous glutamate and BOLD-fMRI signals in the human brain, NeuroImage, 155, 113-119, https://
doi.org/10.1016/j.neuroimage.2017.04.030.
31. Taylor, R., Neufeld, R. W. J., Schaefer, B., Densmore, M., Rajakumar, N., Osuch, E. A., Williamson, P. C., and
Théberge, J. (2015) Functional magnetic resonance spectroscopy of glutamate in schizophrenia and major depres-
sive disorder: anterior cingulate activity during a color-word Stroop task, NPJ Schizophr., 1, 15028, https://doi.org/
10.1038/npjschz.2015.28.
32. Willard, S. S., and Koochekpour, S. (2013) Glutamate, glutamate receptors, and downstream signaling pathways,
Int. J. Biol. Sci., 9, 948-959, https://doi.org/10.7150/ijbs.6426.
33. Gsell, W., Burke, M., Wiedermann, D., Bonvento, G., Silva, A. C., Dauphin, F., Bührle, C., Hoehn, M., and
Schwindt, W. (2006) Differential effects of NMDA and AMPA glutamate receptors on functional magnetic res-
onance imaging signals and evoked neuronal activity during forepaw stimulation of the rat, J. Neurosci., 26,
8409-8416, https://doi.org/10.1523/JNEUROSCI.4615-05.2006.
Publisher’s Note. Pleiades Publishing remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations. AI tools may have been used in the translation or editing of this article.
