KKKeiichi Kitajo

Keiichi Kitajo

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Keiichi Kitajo has a total of 93 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Asymmetric neural entrainment at resonance frequencies underlies unilateral spatial neglect

    by Yuka O Okazaki, Noriaki Hattori, Teiji Kawano, Megumi Hatakenaka, Ichiro Miyai, Keiichi Kitajo

    Unilateral spatial neglect (USN) is a common consequence of right-hemisphere stroke, traditionally attributed to structural lesions and dysfunctional attention networks. However, the brain is fundamentally a rhythmic and dynamical system, and how disrupted neural synchronization underlies USN remains unknown. We recorded steady-state visual evoked potentials (SSVEPs; 3 - 30 Hz flicker) in stroke patients with USN, without USN, and healthy controls. Only the USN group exhibited significant hemispheric asymmetry at 9 Hz, driven by exaggerated responses in the intact hemisphere rather than suppression in the lesioned hemisphere. This effect appeared only during stimulation, not at rest, indicating its specificity to sensory processing. The enhanced 9 Hz entrainment in the intact hemisphere was accompanied by increased phase-amplitude coupling (PAC) between alpha phase and gamma amplitude, reflecting systematic coordination of high-frequency activity. Transfer entropy analysis further revealed increased feedforward information flow from the right visual to the left frontal cortex, highlighting large-scale asymmetry. To explore the mechanism underlying this frequency-specific bias, we implemented a coupled-oscillator model. The model showed that the hemispheric asymmetry arises from resonance between intrinsic alpha rhythms and external input, amplified by asymmetric right-to-left interhemispheric coupling. These findings suggest that USN arises from a selective impairment of alpha-band synchrony capacity. This study offers a novel framework conceptualizing USN as a disorder of disrupted oscillatory dynamics underlying spatial attention, and points toward frequency-specific neuromodulatory intervention as a potential therapeutic approach.

  • Open Access

    Distinct roles of brain network flexibility in motor learning across age

    by Kazumasa Uehara, Makoto Hagihara, Keiichi Kitajo

    Motor learning is a lifelong process, from infancy through old age. Acquiring new motor actions through repetitive practice requires adjusting motor output in response to sensory input and integrating them to facilitate learning. For this to occur, the central nervous system must flexibly predict and adapt to the dynamic interplay between sensory inputs and motor outputs. Although overall brain function changes with age, it remains unclear how flexible brain networks, reflecting moment-to-moment reconfiguration of large-scale brain networks, influence motor learning ability with aging. To address this, we designed a visuomotor learning task involving both younger and older adults and quantitatively assessed brain network flexibility, leveraging multichannel electroencephalography (EEG) in humans. We found age-group differences in motor learning properties, brain network flexibility, and their neural relationships. In younger adults, the learning aftereffect was associated with brain network flexibility during motor preparation in the learning task. However, this association was not observed in older adults. Together, our findings suggest that brain network flexibility during motor preparation was critical for acquiring and maintaining new motor actions in younger adults, but that this coupling was attenuated with aging.

  • Open Access

    The functional significance of EEG phase synchronization networks during information integration of left and right visual fields

    by Makoto Hagihara, Kazumasa Uehara, Yuka O Okazaki, Keiichi Kitajo

    Objects moving between the left and right visual hemifields are naturally perceived as continuous entities, although early visual processing independently transmits information from the two hemifields. Therefore, interhemispheric integration of visual information is essential for maintaining an object's identity. Additionally, brain function is thought to be maintained through a dynamic balance between integration and segregation. In this study, we investigated the functional neural architecture underlying visual hemifield integration in healthy adults, using electroencephalography (EEG) and a visual integration task. To capture neural oscillatory networks without relying on prior assumptions regarding electrode pairs or frequency bands, we applied a frequency-inclusive, data-driven network analysis based on an extended network-based statistic. This analysis identified a broadband EEG phase synchronization network that emerged specifically under task conditions with high interhemispheric integration demands. Furthermore, individual differences in behavioral performance were associated with modulation of interhemispheric synchronization, with this relationship differing according to participants' relative performance across task conditions. These findings suggest that visual hemifield integration is supported by large-scale phase synchronization networks spanning multiple frequencies and are consistent with the importance of a balance between integration and segregation.

  • Open Access

    Metastability in EEG phase synchronization networks is associated with autistic traits in a neurotypical cohort

    by Mebuki Izumiya, Yuka O Okazaki, Keiichi Kitajo

    Metastability is a fundamental dynamical property of large-scale brain networks and reflects the capacity of the brain to flexibly reorganize transient coordination patterns. In this study, we investigated whether metastable properties of resting-state electroencephalographic (EEG) phase synchronization networks are associated with individual differences in autistic traits. Resting-state EEG data from 88 neurotypical adults were analyzed using two complementary metrics: synchrony coalition entropy (SCE), which quantifies the diversity of transient phase synchronization patterns, and the metastability index (MSI), which quantifies temporal variance in global phase synchronization. SCE showed frequency-specific associations with the Autism-Spectrum Quotient (AQ) attention-switching subscore at 18–24 Hz and the communication subscore at 4–8 Hz, suggesting that frequency– and network-specific patterns of metastable synchronization are associated with distinct aspects of autistic traits. In contrast, MSI showed a modest association with the social-skill subscore in the lower-beta range, but this effect did not survive a cluster-based permutation test. This exploratory observation suggests that global synchronization variability may capture a weaker, complementary aspect of trait-related metastable dynamics. These findings suggest that, within a neurotypical population, individual differences in autistic traits may be more sensitively captured by the repertoire of transient phase synchronization patterns, as indexed by SCE, than by global phase synchronization variability, as indexed by MSI. Moreover, the associations of distinct AQ subscores with SCE in different frequency ranges suggest that different dimensions of autistic traits may be related to metastable network dynamics operating at different temporal scales. Author Summary: The brain constantly coordinates activity across many regions, and this coordination changes over time rather than remaining constant. Understanding these dynamic patterns is important for explaining individual differences in cognition and behavior. In this study, we focused on a dynamical property called “metastability,” which describes how brain activity flexibly shifts between different patterns of coordination. Instead of remaining in a stable state, the brain repeatedly forms and dissolves coordinated activity across regions. We analyzed brain signals recorded with resting-state electroencephalography (EEG) and examined whether these dynamic patterns were related to individual differences in autistic traits. We found that different aspects of time-varying coordination were linked to different dimensions of autistic traits in a neurotypical population. These findings suggest that examining how brain activity changes over time, rather than relying only on time-averaged measures, can reveal neural features associated with individual differences in autistic traits. Our study highlights metastability as a useful concept for understanding the flexible and dynamic nature of human brain function.

  • Open Access

    Pretraining Objective Shapes Cross-Category Generalization in Affective Image Prediction: A Geometric Analysis of Vision Transformer Encoders

    by Shohei Tsuchimoto, Yuka O. Okazaki, Kenichi Yuasa, Sakura Nishijima, Mebuki Izumiya, Makoto Hagihara, Ryo Fujihira, Keiichi Kitajo

    The geometry of representations learned by deep neural networks is shaped jointly by architecture and pretraining objective, yet disentangling these two factors remains difficult. Here we isolate the contribution of pretraining objective by comparing two Vision Transformers from the same backbone family but trained under different objectives: language–image contrastive learning (CLIP) and ImageNet-21k classification. Using continuous Valence–Arousal prediction on the OASIS dataset as a probe of representational quality, we evaluated frozen features under Leave-One-Theme-Out and Leave-One-Category-Out cross-validation, the latter requiring extrapolation to entirely unseen semantic categories. The contrastively pretrained encoder generalized substantially better than the classification-pretrained encoder under both protocols, with the gap widening sharply when held-out categories required cross-category generalization. To characterize why the two representations differ, we developed a geometric analysis of prediction errors, treating per-image errors as vectors in the affective plane and quantifying their spatial structure via weighted phase-locking, trajectory-based occupancy entropy, and effective dimensionality. The classification-pretrained representation collapsed errors into a small number of attractor regions with a strong center-ward pull, whereas the language-aligned representation distributed errors broadly across the affective space. Layer-wise linear probing further revealed that affective information was distributed across depth in the contrastive encoder but increasingly concentrated in deeper layers of the classification encoder, mirroring the texture-bias and category-anchored statistics characteristic of ImageNet-trained representations. These results provide a representation-geometric account of how the choice of pretraining objective, holding architecture constant, determines whether learned features generalize across semantic boundaries or remain confined to category-bound visual regularities.

  • Open Access

    Validation of an electroencephalography data assimilation-based computational approach for estimating cortical excitation-inhibition balance

    by Hiroshi Yokoyama, Yoshihiro Noda, Masataka Wada, Mayuko Takano, Keiichi Kitajo

    The excitation and inhibition (E/I) balance of neural circuits is a crucial index of neurophysiological homeostasis associated with healthy brain functioning. Although several cutting-edge methods exist to assess E/I balance in an intact brain, they have inherent limitations, such as difficulties in tracking changes in E/I balance over time. To address this, we introduced neural-mass-model-based tracking using a data assimilation (DA) approach. While we previously demonstrated that sleep-dependent E/I changes could be estimated from electroencephalography (EEG) data, the neurophysiological validity of this method had not been directly evaluated. In this study, we developed an enhanced DA-based method and compared its E/I estimates with the concurrent transcranial magnetic stimulation and electroencephalography (TMS-EEG) based measures. Our results revealed significant correlations between the DA-based estimates and TMS-EEG indices of E/I balance in the dorsolateral prefrontal cortex. These findings indicate that our computational approach provides neurophysiologically valid estimations of time-varying E/I balance.


  • Electroencephalographic Phase Synchrony Index as a Biomarker of Post-Stroke Aphasia Recovery

    by Teiji Kawano, Noriaki Hattori, Yutaka Uno, Megumi Hatakenaka, Hajime Yagura, Hiroaki Fujimoto, Keiichi Kitajo, Ichiro Miyai

    Background: Post-stroke aphasia is considered a language network disorder, and neuroimaging may help understand network alterations. However, the prediction of aphasia recovery remains challenging. Objective: We aimed to explore biomarkers for aphasia recovery using a novel clinically feasible method, which we previously reported as useful for evaluating motor recovery, that included the phase synchrony index (PSI) obtained from resting-state 19-channel electroencephalography. Methods: This longitudinal observational study included patients with left frontal ischemic lesions admitted for post-acute rehabilitation. We recorded electroencephalograms at the time of admission. Recovery was defined as a change in composite speech score. Based on electrode settings, we focused on 4 language-related networks: (1) left front-temporal, (2) right front-temporal, (3) inter-frontal, and (4) inter-temporal networks. We first evaluated the correlation between these network PSIs and recovery scores and then the predictive potential of our method using the receiver operating characteristic curve and multivariable regression analyses. Results: We enrolled 24 patients. Electroencephalograms were recorded for a median of 37.0 days after the stroke. The median speech therapy time was 53.7 hours. Inter-temporal PSI (gamma band) was significantly positively correlated with recovery scores (ρ = .642; 95% confidence interval = 0.311-0.834; P = .017). The PSI could predict patients with good recovery (sensitivity = 84.6%; specificity = 90.9%), and the inter-temporal PSIs were useful in predicting recovery (adjusted R 2 = .545). Conclusions: Our results revealed an association between the posterior language network adaptive response and speech recovery in patients with frontal lesions. The PSI may reflect post-stroke network alterations and may be a biomarker of aphasia recovery.

  • Open Access

    Numerical validation of 3D-VAR data assimilation for estimating network dynamics in multivariate EEGs

    by Hiroshi Yokoyama, Keiichi Kitajo

    Data assimilation (DA) is widely used as a data-driven system identification method, enhancing our understanding of the generative mechanisms in the observational data based on the identified model and its parameters. Due to this advantage, DA has gradually begun to be applied in neuroscience studies. Recently, we introduced a DA-based method for reconstructing the internal state of the brain using the neural mass model and human scalp electroencephalography (EEG). This method allows us to estimate the balance of synaptic interactions between excitatory and inhibitory neurons (E/I balance) in the brain only from observed EEGs. Although we confirmed the neurophysiological validity of the proposed EEG-DA method in our recent works, this method cannot be applied to multivariate EEGs to parallelly infer the E/I balance changes and underlying network dynamics. In this study, to address this issue, we proposed an extended version of our DA method specified to multivariate observed EEGs. This method enables us to estimate sensor-level functional network and E/I balance changes of each EEG sensor in parallel. The method was validated by showing that it could parallelly estimate sensor-level network structures and E/I balance changes from synthetic multivariate EEGs. The results of this study indicate that it has the potential to quantify how the E/I balance changes functionally affect the brain network dynamics.