To further understand the neural basis of brain functions, more knowledge is required about the spatiotemporal aspects of information processing by utilizing multimodal brain imaging beyond what we can learn from a single modality. Scalp electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) are the two noninvasive neuroimaging techniques that are commonly used in human neuroscience. In view of their respective strengths and weaknesses being complementary in temporal and spatial aspects, simultaneous EEG-fMRI fusion provides a new means of uncovering the important spatiotemporal information of brain function. This review will begin with the basic measurement principles of EEG and fMRI and the motivation of simultaneous EEG-fMRI fusion. Next, the state of the art in fuzing the simultaneous EEG-fMRI within the context of spatial and temporal approaches will be considered. Finally, cutting-edge simultaneous EEG-fMRI and application trends that hint at future developments are then presented. It is suggested that simultaneous EEG-fMRI is a promising noninvasive tool that can yield insights into the spatiotemporal information of brain function and dysfunction.
Background Brain–heart interaction is a bidirectional regulatory system integrating neural, biochemical, immune, and energy mechanisms. It plays a central role in maintaining homeostasis and is critically involved in brain–heart comorbidity. Methods: Emerging evidence has elucidated that the brain modulates cardiac rhythm, conduction, and contraction via the central autonomic network, while the heart reciprocally influences brain function, emotion, and cognition through vagal afferents, interoceptive pathways, and immune signaling. The newly proposed brain–heart–immune circuit further expands our understanding of diseases such as myocardial infarction and heart failure. Concurrently, technological innovations—including vagus nerve stimulation, transcranial magnetic stimulation, deep brain stimulation, and wireless self‑powered devices—are driving a paradigm shift from empirical intervention toward closed‑loop intelligent regulation.Results and Conclusion Future research should prioritize cross‑disciplinary integration, multimodal signal fusion, and individualized precision intervention, with the goal of advancing mechanism‑driven diagnosis and treatment of brain–heart comorbidity.
Objective In this paper, we propose Ghost-LENet, a lightweight convolutional neural network based on the LENet framework for electroencephalogram (EEG)-based brain–computer interfaces (BCIs), particularly for motor imagery (MI) classification tasks.Methods The proposed model integrates several lightweight design components to improve feature representation while maintaining a very small parameter scale. Specifically, dilated temporal convolutions and stationary wavelet transform are combined in the initial block to capture multi-scale temporal characteristics of EEG signals. In addition, a dynamic residual fusion mechanism with Efficient Channel Attention (ECA), referred to as DR-ECA, is introduced to adaptively balance attention-enhanced features and original features. Furthermore, a Ghost module is incorporated to improve parameter efficiency while preserving feature extraction capability.Results Experimental results on multiple public EEG datasets show that Ghost-LENet achieves competitive classification performance with only a few thousand trainable parameters across datasets. Specifically, Ghost-LENet achieves classification accuracies of 82.18% and 83.05% on the BCI Competition IV-2a and IV-2b datasets, respectively.Conclusion Overall, Ghost-LENet provides a lightweight and efficient decoding framework for EEG-based motor imagery classification, showing a favorable balance between model complexity and classification performance.
The integration of artificial intelligence (AI) into mental health and psychiatry is transforming the diagnostics and treatment of mental disorders, including major depressive disorder (MDD). While the initial and promising applications span diagnostic screening and theraputic chatbots, these first-wave technologies do not directly target the underlying brain alterations or account for the treatment of MDD. Non-invasive brain stimulation (NIBS) holds a tremendous promise to address the biological heterogeneity of MDD, but is currently hindered by highly variable outcomes. Therefore, we posit that synergistic integration of AI with NIBS represents the most promising path to address these difficulties. Importantly, the frontier for AI in depression treatment lies in a paradigm shift: from empirical trial-and-error to data-driven, personalized precision interventions. We argue for a paradigm shift away from AI roles in mental health (e.g., chatbots, diagnostics) toward its deep integration as the core engine for personalized, circuit-based neuromodulation. We highlight the key opportunities this fusion creates: identifying patient-specific neural targets through predictive modeling, developing adaptive closed-loop therapies that respond to real-time brain states, and brain "digital twins" for in silico simulation and protocol optimization. While significant challenges in data standardization, model interpretability, and clinical validation remain, the fusion of AI and NIBS heralds an era of psychiatry that is predictive, personalized, and precise.
Background Transcranial electrical stimulation (tES) is a promising noninvasive neuromodulation technique for home-based sleep intervention, yet traditional long-duration, high-current protocols cause discomfort and low compliance.Methods This study explored the regulatory effects and optimization potential of intermittent short-duration tES on sleep quality. Twenty-two healthy adults participated in three nocturnal sessions design: Sham, low-current stimulation (Lc-Stim), and high-current stimulation (Hc-Stim) conditions. Data were collected via polysomnography (PSG), and sleep stages were scored according to the American Academy of Sleep Medicine (AASM) criteria.Results The results show that intermittent short-duration tES did not significantly increase total sleep time or sleep efficiency (SE) but did significantly reduce sleep onset latency (SOL) in both the Lc-Stim and Hc-Stim groups. Moreover, the intervention facilitated the N2–N3 transition and attenuated sleep fragmentation. At the micro structure level, SO-spindle coupling density was higher in the Hc-Stim group than in Sham. Compared with conventional long-duration and Hc-Stim, scalp discomfort was significantly reduced in both stimulation groups. Additionally, stimulation effects exhibited interindividual variability, with Hc-Stim demonstrating an intensity-dependent threshold effect on micro structure characteristics.Conclusion The short-duration intermittent tES improves sleep quality safely and acceptably by reducing SOL, increasing deep sleep proportion, and enhancing sleep continuity.
Purpose This article pursues three goals: to clarify the conceptual distinction between ontological, functional, and instrumental modeling claims about mental unity and multiplicity; to situate parts-based psychotherapy within contemporary network neuroscience rather than against an outdated unity-versus-multiplicity binary; and to propose an internal family systems (IFS)-informed artificial intelligence (AI) system design for therapist training and psychoeducational support.Materials and methods Conceptual and narrative review drawing on philosophy of mind, contemporary network neuroscience, experimental psychology, and clinical psychotherapy literature. Philosophical positions from Descartes through contemporary analytic and phenomenological accounts are evaluated, followed by systematic review of network neuroscience findings on integration-segregation dynamics, split-brain and dissociation research, and experimental psychology evidence on pluralistic self-modeling. IFS is compared against alternative parts-based frameworks on criteria of structural specificity, relational articulation, and computational tractability.Results The unity-multiplicity binary is shown to be unproductive for both philosophy and neuroscience. Contemporary network neuroscience characterizes cognition as emerging from flexible integration-segregation dynamics among distributed brain networks, framing functional multiplicity as the normal operating condition of mind rather than an anomaly. Among available parts-based frameworks, IFS is identified as the most tractable instrumental modeling template, distinguished by its discrete role taxonomy, formally specifiable interaction dynamics including polarization, blending, and burden-carrying, and its coordinative principle of self-leadership. A three-layer computational design is proposed: a parts layer implementing semi-autonomous dialogue agents with defined activation conditions and behavioral constraints; a self-like integrative layer enforcing coordinative and ethical constraints via constitutional principles; and a safety layer providing crisis detection, boundary maintenance, and prohibition of deep trauma processing. Two target applications are specified and distinguished: structured simulation for therapist training and psychoeducational self-reflection support.Conclusions The mind is best understood as functional multiplicity with emergent, relational unity, in which coordination rather than structure is the source of psychological coherence. IFS provides a structurally clear and relationally articulate template for AI-supported therapist training and psychoeducational tools. The proposed system design is conceptual rather than implemented, and all reported performance illustrations were generated under controlled mock conditions rather than with production language models. Empirical validation, continued growth of the IFS evidence base, and interdisciplinary collaboration among clinicians, philosophers, AI researchers, and ethicists are required before any deployment.
This paper presents a novel approach that utilizes Field-Programmable Gate Array (FPGA) to perform high-precision modeling and real-time simulation of Purkinje cells. Purkinje cells, as the most dominant inhibitory neurons in the cerebellum, play a crucial role in motor control, learning, and memory. However, due to their complex structure and dynamic characteristics, traditional simulation methods struggle to simultaneously meet the requirements of efficiency and accuracy. To address this challenge issue, (1) we designs and implements a Purkinje cell model based on the FPGA platform, which can operate at speeds close to those in vivo and exhibits good scalability. According to Purkinje cells ion channels and synaptic plasticity mechanisms, a systematic analysis of their electrophysiological characteristics and dynamic response behaviors is conducted. Furthermore, a Hodgkin-Huxley computational model is constructed and optimized, enabling real-time computation of cell firing patterns, signal integration, and synaptic weight changes on the FPGA platform. (2) Combining Hodgkin-Huxley-type equations, parameter fitting is performed for major currents such as sodium, potassium, and calcium ions, analyzing the impact of different channel conductance changes on firing patterns. The results show that the model can effectively exhibit new firing patterns of simple and complex spikes, not only accurately verifying the generation mechanisms of simple and complex spikes but also enhancing modeling speed and scalability. This work provides a computational foundation for understanding the information processing mechanisms of the cerebellum, offers a reference for analyzing pathological neural activities, and has positive implications for the development of brain-inspired computing chips.
Background Spike sorting, which isolates single neuron activities from multi-unit neural recordings, plays a pivotal role in neuroscience research. Recently, advanced recording technologies have enabled large-scale neural signal recording involving thousands of channels synchronously. This advancement has led to a growing demand for computationally efficient spike sorting approaches, particularly for brain-implantable devices where power and memory resources are highly constrained.Methods We propose QuanSort, a spiking neural network (SNN)-based spike sorter designed for energy-efficient on-chip computation with neuromorphic chips. It employs a novel "reset-to-mod" membrane potential reset mechanism. This strategy utilizes suprathreshold states during computation, which preserves higher precision in the value quantization process during on-chip deployment while theoretically ensuring the bounded accumulation of quantization error under the Weyl discrepancy norm.Results Experimental evaluations conducted on both synthetic and real-world datasets demonstrate that QuanSort can be effectively quantized to 16-bit integers. Despite this aggressive quantization, the model maintains high spike sorting performance, validating its suitability for resource-constrained environments.Conclusion QuanSort offers a robust and energy-efficient solution for large-scale spike sorting on neuromorphic hardware. By combining the reset-to-mod mechanism with SNN architecture, it successfully balances computational precision with the strict power and memory limitations inherent to implantable brain-computer interfaces.
With the fast-paced life of modern society and the increasing competitive pressure, the incidence of mental (depression, anxiety, and insomnia disorders, etc.) and neurological (Alzheimer’s disease, etc.) diseases is gradually rising, which has become one of the global public health problems. Moreover, these types of diseases have brought enormous troubles to patients’ lives and work, as well as significant burdens to families and society. Growing studies have demonstrated that Transcranial Magnetic Stimulation (TMS), as a novel, noninvasive non-drug treatment tool with transient adverse effects and few complications, could bring new ideas and schemes for the treatment of mental and neurological diseases. Hence, this paper systematically reviews the pathogenesis of insomnia disorder, depression, anxiety, and Alzheimer’s disease, and then the effects and potential mechanisms of TMS in the treatment of these diseases are systematically summarized. The findings can provide an important plan and reference for the clinical treatment and research of these diseases.
Transcranial magnetic stimulation (TMS) has evolved from a focal brain stimulation method to a network-level neuromodulation tool. This review examines the emerging field of multi-site TMS—approaches that target two or more brain regions to influence inter-regional dynamics and large-scale networks underlying cognition and clinical disorders. Three major forms are described: sequential multi-site TMS, which stimulates different targets in succession and is applied in disorders such as depression and Alzheimer’s disease; dual-site TMS, exemplified by cortico-cortical paired associative stimulation (ccPAS), which employs precisely timed pulses to alter directional connectivity and synaptic plasticity; and multi-locus TMS, a hardware-based method enabling simultaneous, electronically guided stimulation of distributed cortical areas with millisecond precision. While each offers distinct advantages for probing and modulating brain networks, challenges remain, such as parameter variability, individual differences, and technical complexity. Adaptive, personalized protocols guided by neuroimaging are needed to address anatomical and functional variability. In parallel, unified computational models are being developed to optimize protocol design and improve reproducibility. Overall, multi-site TMS represents a promising frontier in neuroscience and neurotherapeutics, bridging local stimulation with global brain dynamics to enable more effective and individualized interventions.
Background Distinguishing Alzheimer’s disease (AD) and mild cognitive impairment (MCI) through electroencephalography (EEG) remains challenging due to subtle neurophysiological differences.Method This study establishes an AI-enhanced EEG framework integrating Random Forest classification with sLORETA/Brainstorm source localization to differentiate AD, MCI, and healthy controls (HC) using 80 age-matched participants.Results Our model achieved significant diagnostic accuracy improvements by automating statistically valid data screening (p ≤ 0.05), increasing recognition rates to 93% for AD (vs. 48% manual) and 91% for MCI (vs. 44%). Critically, spectral and spatial analysis revealed distinct neurophysiological signatures: MCI exhibited prefrontal hyperactivity in the β3 frequency band, while AD showed temporal lobe hyperactivation in the β2 band, with both abnormalities preceding structural atrophy.Conclusion These spatial-spectral biomarkers reflect stage-specific pathophysiology with prefrontal network overcompensation in MCI and pathological temporal synchronization in AD, thus providing a noninvasive tool for early detection.
Background Neurodegenerative diseases such as Alzheimer’s Disease (AD) and Mild Cognitive Impairment (MCI) often manifest olfactory dysfunction before overt cognitive symptoms emerge. Olfactory event-related potentials (ERPs) captured via electroencephalography (EEG) may serve as early biomarkers for these conditions.Objective This study proposes a deep learning-based framework integrating olfactory ERP features and Transformer neural networks to classify normal cognition, MCI, and AD.Methods EEG data were collected from 35 participants during an olfactory oddball task, encompassing standard and deviant odor trials. Preprocessing included bandpass filtering (1—40 Hz) and z-score normalization. ERP features, time-frequency components, and spatial activation patterns were extracted. A Transformer-based model was developed to capture temporal dependencies and classify participants across three diagnostic categories.Results The proposed model achieved a classification accuracy of 87% and a macro-averaged F1-score of 0.88 on the validation dataset, outperforming conventional deep learning methods. Notably, the model demonstrated high sensitivity for MCI detection and exhibited strong interpretability, with attention concentrated on time intervals corresponding to established ERP components.Conclusion This study demonstrates the diagnostic potential of Transformer-based models in detecting early neurodegenerative changes using olfactory EEG signals. The proposed framework offers a noninvasive, cost-effective approach to aid in the early diagnosis of MCI and AD, with potential applications in clinical neurodiagnostics.
Large-scale, high-density EEG datasets offer significant potential to uncover the neural mechanisms of disease and advance clinical applications, yet they remain scarce. Here, we present a dataset of 64-channel resting-state EEG recordings comprising 8,416 resting-state EEG recordings from 8,132 participants, encompassing a wide range of 15 neurological and neuropsychiatric disorder categories. Along with EEG recordings, the dataset includes demographic characteristics, diagnostic classifications, and neuropsychological assessment scores. All data underwent rigorous quality control and were fully de-identified in accordance with ethical and regulatory guidelines. This resource provides an unprecedented opportunity for subgroup and cross-disease comparisons, and the development of robust machine learning models for clinical EEG interpretation. Access to the dataset is managed under institutional agreements ensuring privacy and ethical compliance, supporting both fundamental research and translational applications in neurology and psychiatry.
Aim Intermittent theta burst stimulation (iTBS) shows promise in alleviating cognitive dysfunction in neurodegenerative diseases. Pre-treatment with iTBS could potentially modulate the threshold of synaptic plasticity, which may enhanc the facilitative effect of subsequent iTBS sessions. This study aims to investigate the effects of priming iTBS on working memory ability and neural oscillations of rats with Alzheimer’s desease (AD).Methods Thirty-six AD rats were randomly divided into one of three groups: priming iTBS, iTBS and sham stimulation. Phase-amplitude coupling (PAC) and its directionality were analyzed using local field potentials recorded during a working memory (WM) task. Additionally, hippocampal Aβ concentrations were analyzed across different stimulation sessions.Results Compared to the AD group, the iTBS group learned the WM behavioral task rules in a shorter period and exhibited increased modulation index values for theta-gamma coupling between the hippocampus (HPC) and prefrontal cortex (PFC). Notably, iTBS reversed the direction of PAC observed in AD rats and significantly reduced hippocampal Aβ concentration. Furthermore, compared to the use of iTBS alone, the iTBS priming stimulus had a significant effect on improving the phase-amplitude coupling between the PFC theta band and the HPC gamma band, although no observable behavioral differences were detected.Conclusion Both priming and non-priming iTBS are both superior to sham stimulation in improving cognitive performance of AD rats by enhancing cross-frequency PAC between the PFC and HPC. Furthermore, these interventions altered the directionality of PAC, suggesting modified network connectivity. While priming iTBS induced unique effects on neural oscillation coupling strength, these differences did not result in significant behavioral changes.
Background Selective attention is crucial for filtering relevant information, and its impairment is associated with various disorders. Action video games (AVGs) have been suggested to enhance selective attention, but their long-term effects and therapeutic potential remain understudied.Methods Participants were randomly assigned to either the AVG or the control group and underwent an 8-month training program with five weekly sessions. The correlations between functional connectivity (FC) changes and reaction time (RT) improvement were analyzed after training. To predict and evaluate the effect of subjects on digital therapeutics before training, the participants were categorized into low-, medium-, and high-performance groups based on RT improvement and further classified using pre-training EEG data with support vector machine (SVM) and k-nearest neighbors (KNN) models.Results The AVG group showed significantly greater RT improvement compared to controls. Reduced FC between the frontal and parietal lobes was associated with RT improvement. SVM and KNN classifiers achieved around 80% accuracy and AUC values of 0.87 and 0.90, effectively predicting training outcomes.Conclusion This study demonstrates the effectiveness of AVGs in enhancing selective attention, highlighting their potential as digital therapeutics for cognitive enhancement.
Background: Acoustic therapy modulates neural oscillatory rhythms in various brain regions, thereby alleviating symptoms associated with tinnitus. Neural oscillation serves as an efficient mechanism in the brain, synchronizing information and coordinating temporal processes to facilitate effective integration and processing of information through a combination of harmonic and relaxation oscillations. The presence of 1/f noise, however, can obscure the detection of oscillations and diminish their biological interpretability, thereby impeding the investigation of neuroplasticity in tinnitus patients during acoustic therapy. Methods: This study proposes extracting the periodic components based on neural power spectra decomposition to accurately capture narrow-band activities. Subsequently, we investigated the neuroplasticity changes manifested by periodic oscillations during a 75-day long-term acoustic intervention. Results: The results indicate that the periodic component was more sensitive in capturing treatment-induced neural remodelling. The periodic components in alpha/beta progressively diminished, with their observed prominence becoming negligible by the end of the treatment, a trend consistent with the improvement of clinical symptoms. Changes in periodic components exhibited a stronger linear correlation with THI decline, and the interindividual agreement was greater than that of PSD. Conclusion: This study provides novel insights into the impact of long-term acoustic intervention on neural oscillations in tinnitus patients.
Action video gaming (AVG) experience has been linked to cognitive development and brain plasticity, but its impact on the functional plasticity of cerebellum remains unclear. This study examined whether cerebellar functional plasticity was evident in AVG experts after a year-long reduction in gaming time, and whether the effects varied across cerebellar regions. AVG experts and non-experts underwent resting-state fMRI scans at the beginning of the study and were then instructed to limit their AVG time to a maximum of three hours per week for one year. Resting-state fMRI data were collected again at the end of the study. Results showed that, at the end of the study, AVG experts exhibited a decrease in amplitude of low frequency fluctuation (ALFF) in the right lobule VIII, possibly indicating decreased sensorimotor skills, and an increase in ALFF in the cerebellar lobule IV/V and left lobule VI, possibly indicating enhanced emotional and language abilities. Non-experts did not show significant changes in ALFF. These findings indicate cerebellar functional plasticity in AVG experts following a one-year reduction in gaming time, with varying effects across cerebellar regions.
Background Mental health in pilots has become a critical factor affecting flight safety in the aviation industry. Pilots face high-pressure work environments and long working hours, which can lead to mental health issues, potentially affecting their performance, impacting aviation safety.Methods We investigated the neural mechanisms underlying the mental health of civil aviation pilots by analyzing the fractional Amplitude of Low Frequency Fluctuations (dfALFF) using functional magnetic resonance imaging (fMRI). A total of 48 pilots and 39 healthy controls were included, and their emotional states were assessed using the Self-Rating Depression Scale (SDS) and Self-Rating Anxiety Scale (SAS). The dfALFF instability in the precuneus of the pilots was calculated and compared with that of the controls to explore the relationship between brain activity and emotional state.Results The results showed significantly higher dfALFF instability values in the precuneus of the pilots than in the control group. Additionally, instability in the precuneus was positively correlated with SDS and SAS scores.Conclusion These findings highlight the potential of precuneus dfALFF instability as a biomarker to assess pilots’ mental health. This study emphasizes the importance of incorporating neural activity indicators into pilot training and mental health assessments to enhance aviation safety and provide timely interventions for mental health issues.
