Objective.Image-guided positioning, or neuronavigation, is critical for precise targeting of transcranial magnetic stimulation (TMS) and other noninvasive brain stimulation. However, existing commercial systems have limitations in flexibility and extensibility for research applications.Approach.We present new open-source software for neuronavigated non-invasive brain stimulation (NaviNIBS) that provides comprehensive functionality for TMS experiments. NaviNIBS supports imaging data import, target planning, head registration, real-time tool tracking, and integration with robotic positioning and electrophysiology systems. Key features include flexible target specification, support for multiple tracking hardware options, refined head registration techniques, and an extensible addon system.Main results.We describe the software architecture, core functionality, characterization of tracking performance, and example applications of NaviNIBS.Significance.By implementing NaviNIBS and sharing it with the research community, we aim to facilitate methodological improvements and novel experimental paradigms in noninvasive brain stimulation research.

Objective.Few-shot steady-state visual evoked potential (SSVEP) detection remains a major challenge in brain-computer interface (BCI) systems, as limited calibration data often leads to degraded performance. This study aims to enhance few-shot SSVEP detection through an effective data augmentation (DA) strategy.Approach.We propose a mixup-based DA method that generates synthetic trials by linearly interpolating between real SSVEP signals extracted using a sliding window strategy. The interpolation weight is optimized by maximizing the similarity between the mixed signal and both the template and reference signals. The augmented data is then used to train spatial filters for improved SSVEP detection.Main results.The proposed method was evaluated on two benchmark SSVEP datasets using task-related component analysis and incorporating neighboring stimuli data as spatial filters. Results demonstrate that the mixup-based augmentation significantly improves detection accuracy under few-shot conditions, outperforming existing augmentation and baseline methods.Significance.The mixup-based method offers an effective and practical solution for enhancing SSVEP decoding with limited data, reducing calibration time, and improving BCI systems' usability in real-world scenarios.

Objective.Temporal interference stimulation (TIS) has recently been introduced for non-invasive deep brain stimulation (NDBS). While numerous studies have highlighted its advantages over conventional technologies, TIS still encounters challenges such as limited resolution and a lack of validation using human-like models. This article introduces an innovative method for NDBS which alleviates the resolution limit.Approach.We utilize as our excitation a 1.5 GHz microwave carrier modulated by a 10 Hz envelope. The microwave carrier enables strong electromagnetic focusing while the envelope triggers neural activity. To form this excitation, two dipole antenna arrays are placed around the head for the generation ofy- andz-directed electric fields (E-field). Current excitations to the antenna arrays are tuned to control (i) theE-field to the desired focality position and (ii) its direction at the focality position. Full-wave simulations with a realistic head model are conducted to demonstrate the method.Main results.In the deep brain region, the cross-sectional focality sizes (75% threshold) are 0.73 cm2, 1.18 cm2and 2.45 cm2in theXOY, YOZandXOZplanes, respectively. The focality is much smaller than previously reported in the conventional method with kHz carrier waves. Further, theE-field direction at the focality can be steered along theyz-plane by adjusting the excitation weights of the antenna arrays. Multiphysics simulations on temperature distribution and specific absorption rate (SAR) show that the maximum temperature increase within a 30-minute stimulation session is 0.76 °C and the maximum SAR1gis 2.70 W kg-1. Both measures are within commonly accepted safe operation ranges.Significance.Compared to conventional TIS methods that utilize kHz carrier signals, our proposed approach achieves drastically improved spatial resolution and enables precise steering of theE-field. The proposed work holds significant potential for clinical applications, offering enhanced resolution and reduced input power for NDBS.

Objective.Optically pumped magnetometer-based magnetoencephalography (OPM-MEG) offers significant advantages over traditional systems based on superconducting quantum interference devices, including flexibility and the ability to record brain activity without cryogenic cooling. However, OPM-MEG is highly susceptible to motion artifacts due to its sensitivity to external magnetic field fluctuations.Approach.To address this challenge, we propose an Adaptive Time-varying (ATH) Homogeneous field correction (HFC) method, which integrates time-varying HFC with adaptive filtering to suppress head motion artifacts. The ATH method estimates background magnetic field components induced by head movements and dynamically adjusts filter parameters to minimize discrepancies between measured signals and predicted background fields.Main results.We evaluated the ATH method through simulation studies and median nerve stimulation OPM-MEG experiments, demonstrating its effectiveness in enhancing signal quality and robustness across various experimental conditions.Significance.ATH offers an effective solution for motion artifact suppression in OPM-MEG systems. Its robustness under diverse conditions supports broader application in research and clinical settings.

Objective.Accurate tracking of brain states during general anesthesia remains challenging due to the complex neurophysiological dynamics involved.Approach.This study developed a thalamo-cortical neural mass model (TC-NMM) and a mean-field model (MFM) incorporating shared thalamic nuclei, both integrated with a particle filtering (PF) algorithm, to characterize consciousness transitions during sevoflurane- and protocol-induced anesthesia. The PF algorithm was employed to dynamically estimate model parameters, including excitatory/inhibitory postsynaptic potential (EPSP/IPSP), and the time constant rate of EPSP/IPSP, along with the coupling coefficients of the thalamic and cortical modules.Main results.The PF-based TC-NMM and MFM accurately tracked frontal data obtained during sevoflurane anesthesia and thalamo-cortical data acquired during protocol-induced anesthesia, respectively. Parameter estimation results revealed that both sevoflurane and protocol anesthesia reduced thalamo-cortical connectivity, with the thalamo-cortical coupling coefficients reliably distinguishing between distinct consciousness states. Notably, the EPSP parameters and coupling coefficients from the TC-NMM hold potential as clinically viable indicators for monitoring anesthesia depth.Significance.These findings not only advance our understanding of anesthetic mechanisms from a model perspective, but also suggest novel, physiologically interpretable indicators for assessing anesthesia depth.

Objective.Functional brain networks (FBN) are important tools for understanding, classifying and analyzing the brain. However, the multi-term features and temporal correlation of individuals are not adequately represented in single-layer and single-scale FBNs, resulting in room for improvement in the classification accuracy and generalizability of FBNs.Approach.Based on the temporal variability and spatial distribution of electroencephalography (EEG), a multi-scale spatio-temporal FBN is constructed on both temporal and spatial scales. Firstly, brain field data aggregation computation. Based on Ising model design the method of brain field data aggregation, represent whole characteristics of brain field with a symbol, and map multiple time series into a symbol sequence. Secondly, autocorrelation calculation between symbol subsequences. Divide sequence into multiple non-overlapping subsequences, compute the autocorrelation between subsequences based on Kronecker Delta, and represent the relationships between the states of the brain over time. Thirdly, spatio-temporal FBN construction. Subsequence are taken as nodes, and symbol sequence correlations are used as link weights, temporal FBN is constructed. Within each node of the temporal FBN, channels are taken as nodes, and functional connectivities of inter-channel time series are used as link weights, spatial FBN is constructed. Finally, the spatio-temporal FBN is applied for EEG classification.Main results. The classification accuracies of the spatio-temporal FBN are up to 99% on fatigue detection, emotion recognition, Parkinson's diagnosis and motor imagery datasets. Thereby, it is verified that the spatio-temporal FBN possesses satisfactory effectiveness, efficiency and generalizability.Significance. The advantages of the spatio-temporal FBN are that the short-term and long-term features of individuals and categories are represented, while enabling universal recognition among different individuals and distinction among different categories.

Objective.Sensory feedback restoration represents a fundamental need for upper limb prosthesis users because it permits to feel somatic sensations during interactions with the environment. Considering the artificial sensory transduction, neuroprotheses should take advantage of effective encoding algorithms, which have the essential role, in the sensory feedback process, of coding the intended perception to the individual with the amputation. This paper presents a literature systematic review of the encoding algorithms employed for somatotopically restoring somatic sensations in upper limb of individuals with the intact arm or with an amputation.Approach.The methodologies adopted for the development of the encoding algorithms were deeply analyzed to describe what is the current state of the art on this topic. Encoding algorithms validated in literature on upper limb were grouped into three main categories (Function-based,Bio-inspiredandHybrid) and then compared and described.Main results.Function-based Algorithmsprovide the user with high sensitivity, whereas if the verisimilitude to natural sensation and complexity are the most desirable features for sensory feedback, aBio-inspiredstrategy would be the most suitable to implement. However,Hybridsolutions both evoked realistic sensations and enhanced discrimination capabilities.Significance.The conducted analysis represents a guide for understanding which type of encoding to choose, making a compromise between the characteristics of the elicited sensations and the achieved performance. This critical analysis will give the reader important information for understanding the potentiality of the encoding strategies to elicit different sensations for a specific application and for developing novel sensory restoration approaches.

Objective. This study aimed to investigate the potential of contrastive learning to improve auditory attention decoding (AAD) using electroencephalography (EEG) data in challenging cocktail-party scenarios with competing speech and background noise.Approach. Three different models were implemented for comparison: a baseline linear model (LM), a non-LM without contrastive learning (NLM), and a non-LM with contrastive learning (NLMwCL). The EEG data and speech envelopes were used to train these models. The NLMwCL model used SigLIP, a variant of CLIP loss, to embed the data. The speech envelopes were reconstructed from the models and compared with the attended and ignored speech envelopes to assess reconstruction accuracy, measured as the correlation between the reconstructed and actual speech envelopes. These reconstruction accuracies were then compared to classify attention. All models were evaluated in 34 listeners with hearing impairment.Results. The reconstruction accuracy for attended and ignored speech, along with attention classification accuracy, was calculated for each model across various time windows. The NLMwCL consistently outperformed the other models in both speech reconstruction and attention classification. For a 3-second time window, the NLMwCL model achieved a mean attended speech reconstruction accuracy of 0.105 and a mean attention classification accuracy of 68.0%, while the NLM model scored 0.096 and 64.4%, and the LM achieved 0.084 and 62.6%, respectively.Significance. These findings demonstrate the promise of contrastive learning in improving AAD and highlight the potential of EEG-based tools for clinical applications, and progress in hearing technology, particularly in the design of new neuro-steered signal processing algorithms.

Objective.Effective characterization of neural complexity during motor execution tasks enhances understanding of maladaptive cortical reorganization in stroke and inform targeted rehabilitation. While traditional EEG analyses often do not consider nonlinear temporal dynamics, we introduce a recurrence based computational framework to quantify cortical complexity during hierarchical motor tasks. Here, we evaluate contralesional motor system engagement in stroke survivors using recurrence quantification analysis (RQA), ensuring sensitivity to nonlinear and temporally structured cortical activity.Approach. RQA was applied to EEG signals recorded during shoulder abduction (SABD) at 20% and 40% torque levels to characterize nonlinear cortical dynamics and quantify complexity distinguishing adaptive from maladaptive motor system engagement. Spatially resolved recurrence metrics were compared between stroke and control participants to elucidate compensatory cortical reorganization linked to motor impairment and hierarchical task demands.Results. Our findings show a statistically significant increase in EEG signal complexity within the contralesional hemisphere of stroke participants, particularly under higher SABD loads. Consistent with previous studies, we observed abnormal muscle coactivation patterns between proximal and distal muscles, along with distinct shifts in EMG vector direction in stroke-impaired limbs. These shifts in coactivation patterns suggest constraints in muscle coactivation patterns resulting from losses in corticofugal projections and upregulated brainstem pathways.Significance. We introduce a novel application of RQA to quantify nonlinear EEG complexity during motor execution in chronic stroke. Our results show that increased EEG complexity reflects greater recruitment of contralesional motor pathways, indicating maladaptive cortical reorganization linked to impaired motor control. Unlike traditional spectral or connectivity-based EEG signal processing methods, RQA quantifies temporally evolving, nonlinear recurrence dynamics, serving as a marker of maladaptive contralesional motor recruitment, positioning RQA as a promising, clinically meaningful, and computationally efficient tool to evaluate cortical dynamics and guide targeted neurorehabilitation strategies aimed at minimizing maladaptive plasticity.

Objective.Accurate localization of the epileptogenic zone (EZ) is crucial for epilepsy surgery, but the class imbalance of epileptogenic vs. non-epileptogenic electrode contacts in intracranial electroencephalography (iEEG) data poses significant challenges for automatic localization methods. This review evaluates methodologies for handling the class imbalance in EZ localization studies that use machine learning (ML).Approach.We systematically reviewed studies employing ML to localize the EZ from iEEG data, focusing on strategies for addressing class imbalance in data handling, algorithm design, and evaluation.Results.Out of 2,128 screened studies, 35 fulfilled the inclusion criteria. Across the studies, the iEEG contacts annotated as epileptogenic prior to automatic localization constituted a median of 18.34% of all contacts. However, many of these studies did not adequately address the class imbalance problem. Techniques such as data resampling and cost-sensitive learning were used to mitigate the class imbalance problem, but the chosen evaluation metrics often failed to account for it.Significance.Class imbalance significantly impacts the reliability of EZ localization models. More comprehensive management and innovative approaches are needed to enhance the robustness and clinical utility of these models. Addressing class imbalance in ML models for EZ localization will improve both the predictive performance and reliability of these models.