Kapitel

• FrontmatterMüller-Putz, Gernot; Kostoglou, Kyriaki; Oberndorfer, Markus E.; Wriessnegger, Selina C.; 10.3217/978-3-99161-093-9-000
• Regulatory challenges of certification of implantable BCI medical devices in the EU and how to overcome themMachado, Ana Matos; 10.3217/978-3-99161-093-9-001Brain Computer Interface (BCI) technology has seen substantial growth over the past decade with increasing private investment into its research and development. With ongoing promising clinical studies, market entry of implantable devices is not far in the future and so regulatory discussions now start to become relevant. All medical devices in the EU fall under the scope of the Medical Device regulation 2017/745 (MDR) and potentially the AI Act 2024/1689. So how will certification work for this technology? What will be the expected challenges considering the current regulatory framework? With focus on the clinical assessment part of certification, this article will try to navigate these questions and provide clarity to the process.
• Sensorimotor EEG Activity During Brain-Computer Interface Therapy Predicts Motor Function in Parkinson’s DiseaseSieghartsleitner, Sebastian; Cho, Woosang; Sebastián-Romagosa, Marc; Ortner, Rupert; Schwarzgruber, Michael; Scharinger, Josef; Guger, Christoph; 10.3217/978-3-99161-093-9-002Parkinson’s disease (PD) is associated with motor impairment and altered sensorimotor oscillatory activity. Brain computer interface (BCI) based neurorehabilitation may provide both therapeutic benefits and a means to assess underlying neurophysiology. In this study, we investigated whether people with PD (pwPD) can effectively use a motor imagery (MI) based BCI and whether MI related neurophysiological markers allow prediction of motor function at the individual subject level. Electroencephalography (EEG) data were analyzed from pwPD performing BCI therapy based on MI, functional electrical stimulation and visual feedback using a 3D avatar. PwPD achieved a mean BCI accuracy of 74%, which was not related to motor symptom severity as quantified by the Movement Disorders Society Unified Parkinson Disease Rating Scale (MDS-UPDRS) Part III. Using cross validated regression models, beta band event related (de)synchronization (ERD/S) during MI of all four limbs reliably predicted MDS-UPDRS Part III on an individual subject level. These findings demonstrate that pwPD can control an MI based BCI independent of motor impairment severity and extend existing literature beyond group level differences by identifying beta ERD/S as an individualized neurophysiological marker of motor function in PD.
• Unsupervised Intent Gating: Improving Calibration and Online Reliability in Motor Imagery BCIsForin, Paolo; Tortora, Stefano; Menegatti, Emanuele; Tonin, Luca; 10.3217/978-3-99161-093-9-003Motor Imagery (MI) Brain-Computer In terfaces (BCIs) are frequently hampered by fluctuations in user attention, which introduce noisy labels during standard calibration. To address this, we propose an unsupervised gating architecture using a Gaussian Mix ture Model (GMM) to dynamically estimate Intentional Control (IC). By evaluating sensorimotor rhythm spa tial focality, the GMM automatically filters out non informative idling segments without manual pruning. Of fline validation (N=5) demonstrated sharper neurophysi ological topographies and higher class separability than standard methods. During real-time closed-loop control, the GMM probability dynamically weighted a continu ous soft-increment integrator. While baseline sample wise accuracy remained comparable (77.7% vs. 77.4%), the gating mechanism effectively converted neural uncer tainty into safe timeouts, boosting effective active deci sion accuracy to 92.7% and maintaining a 76.0% safety margin against false positives during rest. This intent gated paradigm offers a robust framework for safe, real world assistive BCI technologies.
• Confidence-weighted cumulative rCCA with post hoc re-analysis: unsupervised adaptive learning for calibration-free c-VEP BCIThielen, Jordy; 10.3217/978-3-99161-093-9-004Abrain–computerinterface(BCI)typically requires a calibration period to train a decoding model, which is a tedious and time-consuming process. Previous studies have shown high performance of calibration-free decoding approaches, like adaptive reconvolution canon ical correlation analysis (rCCA). rCCA can operate in an instantaneous mode, classifying trials independently without prior information, but can be improved by cumu latively learning from previously classified trials. Such cumulative strategy, however, relies on potentially un reliable pseudo-labels. In this study, two extensions to unsupervised rCCA are introduced. First, confidence weighting ensures that model updates are driven primar ily by high-confidence trials. Second, a post hoc re analysis allows misclassifications to be corrected using a later, more accurate model. Four rCCA variants were evaluated on an existing code-modulated visual evoked potential (c-VEP) dataset, showing no significant effect of confidence-weighting, but a significant improvement of post hoc re-analysis. These findings pave the way for robust and fully calibration-free c-VEP BCI systems.
• Realtime Class Balancing for Efficient Decoder Adaptation in Asynchronous Brain-Computer InterfaceSouriau, Rémi; Martel, Félix; Costecalde, Thomas; Karakas, Serpil; Aksenova, Tetiana; 10.3217/978-3-99161-093-9-005Brain-Computer Interfaces (BCIs) allow direct control of external effectors via brain activity de coding. For real-world use, BCIs must operate in an asyn chronous mode that enables users to initiate or withhold system use, while maintaining low false positive rates (FPR) and high decoding accuracy. Class balance is cru cial for efficient model training. The RSW-NPLS algo rithm has been proposed recently to balance classes dur ing online and incremental learning. However, perfect balancing may not be optimal for asynchronous BCIs, as they require a highly stable idle state. We propose a class balancing strategy that prioritizes the idle state to reduce its false activations of the other states, and evaluated it on four databases from tetraplegic and paraplegic users. It significantly reduced the FPR (between-55.36% and 70.08%) with minimal loss of accuracy (between-3.68% and-10.51%).
• Combining Virtual Reality Feedback and Transcranial Alternating Current Stimulation to Enhance Restorative Brain-Computer InterfacesSaraiva, Joana; Esteves, Daniela; Bour, Christel; Fernandes, Sofia; Vourvopoulos, Athanasios; 10.3217/978-3-99161-093-9-006Instrokerehabilitation, brain–computer in terfaces (BCIs) enable neurofeedback-based regulation of sensorimotor rhythms, promoting motor-related corti cal reorganization. Integrating virtual reality (VR) into BCI paradigms enhances this process through immer sive, multisensory feedback that strengthens sensorimo tor engagement. In parallel, transcranial alternating cur rent stimulation (tACS) may further modulate motor im agery(MI)–related oscillatory activity. Together, these approaches offer a multimodal framework to optimize neurorehabilitation outcomes. This study investigated the synergistic effects of a MI-based VR-BCI paradigm and tACS applied at the individual alpha frequency (IAF). Twelve healthy participants performed bimanual MI tasks in VR and non-VR conditions before and after sham or active tACS. Alpha event-related desynchronization (ERD) served as the primary outcome. VR elicited stronger contralateral and more widespread ERD, includ ing parietal and occipital regions. tACS effects were sub tle and, in VR, sometimes attenuated ERD.
• Efficient Decoding of the Code-Modulated Motion Visual Evoked Potential (c-MVEP) for BCIsScheppink, Hanneke A.; Bialonski, Stephan; Tangermann, Michael; Thielen, Jordy; Volosyak, Ivan; 10.3217/978-3-99161-093-9-007The code-modulated motion visual evoked potential (c-MVEP) paradigm visually stimulates targets using motion in pseudo-random intervals. Due to its code-modulated nature, c-MVEP shares aspects of the code-modulated visual evoked potential (c-VEP). This study investigated whether the c-MVEP can be decoded using similar classification approaches as used for c VEP. Several canonical correlation analysis (CCA)-based methods were evaluated, with increasingly constrained assumptions about the neural data, to reduce model com plexity. These included ensemble and global spatial fil ters, as well as template estimation by averaging across trials, cycles, classes, or stimulus events. Consistent with findings in c-VEP, we found that the event-based ap proaches, with events defined by the rising or falling edge of the stimulus sequence, achieved a higher decoding ac curacy with fewer training data for calibration than the commonly used trial-based approach. This study shows the potential for decoding the c-MVEP with high accu racy, without requiring extensive training.
• Advancing NIRS-Based Insights into Motor Imagery and Execution of Swallowing: Effects of Artifact Correction Methods and Hemodynamic ValidationKober, Silvia Erika; Engele, Veronika Isabella; Kanatschnig, Thomas; 10.3217/978-3-99161-093-9-008Motor imagery (MI) of swallowing has proven to be a promising mental strategy for brain computer interface (BCI) and neurofeedback applications aimed at treating swallowing difficulties. Previous near-infrared spectroscopy (NIRS) studies demonstrated comparable hemodynamic responses over the inferior frontal gyrus during motor execution (ME) and MI of swallowing, particularly for deoxygenated hemoglobin. However, due to the reliance on manual artifact rejection, it remained unclear whether these findings were influenced by motion artifacts. In this study, 33 healthy young adults were tested, and their hemodynamic responses during ME and MI of swallowing were assessed using NIRS. By applying advanced and automated motion artifact correction methods, such as wavelet filtering and short-distance channel regression, the results were largely consistent with previous studies that used manual artifact rejection. This validates earlier findings and confirms that the observed activation patterns in the hemodynamic response were not primarily caused by motion artifacts, demonstrating the robustness of prior conclusions. When directly comparing motion artifact correction methods (manual rejection, wavelet filtering, and short-distance channel regression), wavelet filtering effectively reduced concentration changes in oxygenated hemoglobin compared to manual correction, indicating fewer motion artifacts. In contrast, short-distance channel regression showed no significant effects, underscoring the need for further optimization of correction methods and the importance of selecting appropriate artifact correction techniques in NIRS studies investigating ME and MI of swallowing.
• Inter-Expert Variability in Common Spatial Pattern Component Selection: Implication for Neurofeedback ApplicationsDumas, Cassandra; Dussard, Claire; Corsi, Marie-Constance; George, Nathalie; 10.3217/978-3-99161-093-9-009Common Spatial Patterns (CSP) are widely used in motor imagery (MI)-based brain computer interfaces (BCIs). While classification pipelines combine multiple components to optimize performance, MI-based neurofeedback (MI-NF) requires selecting a single spatial filter to drive feedback based on physiologically interpretable sensorimotor activity. This critical step is often implicit and relies on expert judgment. We quantified inter-expert variability in CSP selection using data from 20 subjects in a right-hand MI NF experiment. Twenty-three BCI or neurophysiology experts independently selected the most physiologically relevant component among six CSP candidates per subject. Inter-expert agreement was fair overall (Fleiss’ κ = 0.256) but varied substantially across subjects. Consensus did not systematically favor the top-ranked component, and selections spanned all six candidates. Although higher-ranked components elicited stronger agreement, substantial variability persisted. These findings identify single-component CSP selection as an overlooked source of methodological heterogeneity in MI-NF systems, highlighting the need for explicit and reproducible selection criteria.
• Resting-State EEG Predictors for BCI Performance: A Scoping ReviewSettgast, Tomko; Patel, Rishan; Kübler, Andrea; 10.3217/978-3-99161-093-9-010This scoping review summarizes studies investigating the association between resting-state EEG and subsequent BCI performance. Following PRISMA ScR guidelines, we systematically mapped the literature to provide a comprehensive overview. We discuss the results within the context of BCI performance prediction, emphasizing the importance of resting-state brain activity, particularly for patient-centered research.
• Comparison of sequence, stimulus and decoding algorithm combinations in c-VEP-based BCIsMartín-Fernández, Ana; Martínez-Cagigal, Víctor; Santamaría-Vázquez, Eduardo; Hornero, Roberto; 10.3217/978-3-99161-093-9-011Brain-computer interfaces (BCIs) enable users to control external devices using brain activity. Among exogenous paradigms, code-modulated visual evoked potentials (c-VEP) stand out. These are neural responses elicited by flickering visual stimuli encoded with pseudorandom sequences, allowing the system to identify the target attended by the user. Optimizing c VEP performance requires understanding the combined effects of encoding sequences, stimulus patterns, and de coding algorithms. In this study, 26 healthy participants evaluated a 16-target speller with 6 c-VEP configurations, combining 2 sequences (m-sequences, burst codes) with 3 stimulus patterns (plain, grating, checkerboard) using 4 decoding algorithms. Results showed that burst codes with a checkerboard pattern and a bit-wise reconstruc tion decoder offered the best balance between accuracy (99.76%) and comfort (eyestrain score 4.81/10). How ever, several combinations also yielded reliable perfor mance, highlighting that optimal design may depend on the specific application needs. This work emphasizes the importance of jointly considering sequence type, stimu lus pattern, and decoding algorithm in c-VEP research.
• Wearable brain-computer interfaces and their applications: advancements from the ARHeMlab research groupEsposito, Antonio; Moccaldi, Nicola; Gargiulo, Ludovica; Duraccio, Luigi; De Benedetto, Egidio; Galasso, Enza; Di Marino, Lucrezia; Angrisani, Leopoldo; Arpaia, Pasquale; 10.3217/978-3-99161-093-9-012This work presents recent advancements in the development of wearable brain–computer interfaces (BCIs) bythe ARHeMlabresearchgroup, focusingonthe period from 2021 to 2025. The activities address the key challenges of translating BCIs from controlled laboratory settings to real-world, daily-life scenarios. The research integrates wearable and portable EEG hardware with cus tom signal processing pipelines and XR environments to enhance signal quality and user engagement. Exper imental contributions span reactive, passive, and active BCI paradigms, including SSVEP-based interfaces with adaptive XR stimulation, cognitive load and distraction monitoring using low-channel EEG, and motor imagery control with multimodal neurofeedback. The proposed solutions are validated through application-specific stud ies, supporting the feasibility of practical, scalable, and ecologically valid BCI systems. No original research is presented in this work but the abovementioned works are reviewed.
• Evaluating Baseline Correction Techniques for EEG Time-Frequency Feature ClassificationHons, Manuel; Kober, Silvia Erika; 10.3217/978-3-99161-093-9-013EEG-based classification is sensitive to preprocessing procedures. Although baseline correction is widely applied in EEG research, its impact on classification performance remains underexplored. We compared four baseline strategies: (I) amplitude domain correction, (II) normalized power-domain correction, (III) combined amplitude- and power domain correction, and (IV) no baseline correction. Analyses were conducted using both an empirical speech imagination dataset and simulated data. Across both datasets, power-domain baseline correction produced significantly lower classification accuracies than amplitude-domain correction and no baseline correction. The replication of this pattern in simulated data indicates that these differences reflect fundamental signal-processing properties rather than dataset-specific confounds. These results suggest that baseline correction applied after nonlinear power transformation can distort feature distributions, whereas amplitude-domain correction removes additive offsets prior to nonlinear expansion, thereby preserving more stable variance structures.
• Riemannian Geometry-Preserving Variational Autoencoder for MI-BCI Data AugmentationPoļaka, Viktorija; De Jong, Ivo Pascal; 10.3217/978-3-99161-093-9-014This paper addresses the challenge of gen erating synthetic electroencephalogram (EEG) covari ance matrices for motor imagery brain-computer inter face (MI-BCI) applications. Objective. We aim to de velop a generative model capable of producing high fidelity synthetic covariance matrices while preserving their symmetric positive-definite nature. Approach. We propose a Riemannian geometry-preserving variational autoencoder (RGP-VAE) integrating geometric mappings with a composite loss function combining Riemannian distance, tangent space reconstruction accuracy and gen erative diversity. Results. The model generates valid, representative EEG covariance matrices, while learning a subject-invariant latent space. Synthetic data proves practically useful for MI-BCI, with its impact depend ing on the paired classifier. Contribution. This work introduces and validates the RGP-VAE as a geometry preserving generative model for EEG covariance matri ces, highlighting its potential for signal privacy, scalabil ity and data augmentation.
• A Feasibility Study of a Motor Imagery–Based Brain–Computer Interface for Post-Stroke PatientsMartínez-Cagigal, Víctor; Moreno-Calderón, Selene; Cisnal, Ana; Pérez-Velasco, Sergio; Martín-Fernández, Ana; Santamaría-Vázquez, Eduardo; Fraile, Juan C.; Pérez-Turiel, Javier; Hornero, Roberto; 10.3217/978-3-99161-093-9-015Strokeisamajorcontributor to chronic dis ability, and many survivors continue to experience pro nounced upper-limb weakness despite standard rehabil itation. By converting motor intent into real-time sen sory feedback, brain–computer interfaces (BCIs) could help re-engage the sensorimotor loop. However, find ings across clinical studies remain mixed, especially for robot-assisted approaches. In this feasibility study, we investigated whether a noninvasive motor-imagery (MI) BCI providing concurrent multimodal feedback (visual avatar and robot-assisted hand motion) yields additional improvements beyond usual care. Sixteen patients with a first-ever stroke were assigned to conventional therapy alone (CG, n = 6) or to the same therapy supplemented with 23 one-hour BCI sessions(EG,n=10). OnlineEEG from eight sensorimotor channels was used to classify left- versus right-hand kinesthetic MI, and motor, cog nitive, and functional tests were evaluated before and af ter the intervention. The EG showed significant within group gains in voluntary motor control and coordination of the affected hand, episodic memory and visuocon structive ability, as well as functional measures such as perceived general health and independence. These pre liminary findings suggest that BCI-based therapies may enhance post-stroke rehabilitation.
• Noninvasive self-paced BCI for lower limb exoskeleton control: a multi-session investigation of user learningFaro, Alessio Lo; Cortecchia, Tommaso; Mihalovic, Miroljub; Trombin, Edoardo; Menegatti, Emanuele; Tonin, Luca; Tortora, Stefano; 10.3217/978-3-99161-093-9-016Brain-Computer Interfaces (BCIs) repre sents a promising approach to incorporate the user’s motion intention into the control of assistive technolo gies. However, its efficacy for the control of real lower limb exoskeletons (LLE) remain largely overlooked, with most of the work in the literature focusing on single session studies. To overcome these limitations, this work proposes and investigates the usage of a noninvasive electroencephalography (EEG)-based BCI for self-paced control of a LLE across multiple sessions. The proposed framework and training protocol are explicitly designed to promote progressive familiarization through repetitive training, continuous feedback, and minimal decoder cal ibration, encouraging the emergence of more stable and physiologically interpretable neural patterns. Experimen tal results show that all participants achieved functional LLE control within three sessions, paired with classifica tion improvements and stabilization of cortical activation patterns. Overall, the findings support the feasibility of practical LLE-BCI systems while highlighting the impor tance of stable decoding strategies in online and multi session experiments promoting user learning, which are key to achieving a more reliable and effective BCI-driven LLE control.
• The Effect of Single Session Implicit Motor Learning on Finger Motor Imagery for EEG BCIDamm, Laura M.; Jiang, Dai; Demostheneous, Andreas; 10.3217/978-3-99161-093-9-017Motor imagery (MI) based brain computer interfaces (BCIs) rely on cue driven paradigms, yet little is known about how the order in which cues are presented influ ences neural discriminability. Here the effect of sequen tial versus random cue order on finger-MI is investigated. Ten healthy, right-handed participants (7 F, mean age = 31.9 ± 10.7 years) completed two runs of 300 trials each (30 cycles of 5-finger trials). Across subjects, the γ band integrals showed the most consistent cue-order ef fect. Depending on frequency band 7-8 subjects yielded significant effects. Participants who began with the se quential cues exhibited higher mean amplitudes and re duced variability across runs, suggesting that early ex posure to a predictable cue sequence facilitates implicit motor-learning and stabilises the forward model. These findings demonstrate that cue-order manipulation is a po tent factor influencing MI-BCI performance and under score the importance of designing training protocols that promote implicit learning.
• Diversity in EEG: Why is it important?Kelly, Merlin Angel; Patel, Rishan; Bryan, Peter; Kier, Lexi; Carlson, Tom; 10.3217/978-3-99161-093-9-018Electroencephalography (EEG)-based brain-computer interfaces (BCIs) are increasingly deployed across clinical and consumer contexts, yet the research foundations underpinning these systems reflect a profound lack of demographic diversity. Hardware limitations, particularly the inability of standard EEG electrode systems to achieve reliable scalp contact across curly, kinky, and afro-textured hair types, represent a concrete and under acknowledged barrier to inclusive data collection. We present findings from a mixed-methods survey of 31 EEG researchers and practitioners, finding that 74% reported hardware-related challenges with diverse hair types, with difficulties clustering at cap placement and preparation. Qualitative themes revealed patterns of signal degradation, extended setup burden, explicit participant exclusion, downstream data loss, and a systemic researcher knowledge gap. Existing hardware solutions only partially address this problem. We argue that electrode redesign capable of achieving reliable scalp contact across all hair types is a prerequisite for BCI systems that work equitably for all.
• Decoding Sleep and Wakefulness from Chronically Implanted Intracortical Signals in a Communication BCI UserOffenberg, Elena Charlotte; Freudenburg, Zachary V.; Branco, Mariana P.; Zimmermann, Jonas B.; da Cruz, Janir R.; Vansteensel, Mariska J.; Ramsey, Nick F.; 10.3217/978-3-99161-093-9-019As brain-computer interfaces (BCIs) move toward long-term home use, continuous operation across day and night introduces new challenges. Decoders trained during wakefulness may produce unintended activations during sleep due to large differences in neural activity between vigilance states. Users must be able to communicate at night, making reliable detection of sleep and wakefulness essential for safe and autonomous BCI operation. We investigated whether awake-sleep stages can be inferred from chronically implanted intracortical neural signals in a communication BCI user with complete locked-in syndrome. Simultaneous non-invasive electroencephalography (EEG) and electrooculography (EOG) recordings provided reference sleep labels over a 44-hour period. A convolutional neural network was trained to distinguish between wakefulness and N2 sleep using intracortical data and achieved high accuracy (96.1±1.6%). The model was applied to six independent BCI spelling sessions, where more sleep-classified epochs coincided with reduced spelling performance. These results demonstrate the feasibility of decoding vigilance state from intracortical signals and support the development of state-aware BCIs for continuous home use.
• Proteus Effect in VR-Based BCI: How Avatar Age Influences Motor Imagery Behavior and EEGEsteves, Daniela; Vasconcelos, Marta; Vourvopoulos, Athanasios; 10.3217/978-3-99161-093-9-020The Proteus Effect suggests that embodying avatars in virtual reality (VR) with specific characteristics can activate related stereotypes and alter behavior. While prior work has shown that avatar age influences motor imagery (MI) performance, its impact on MIrelated EEG activity remains unclear. This pilot study investigated whether avatar age modulates both behavioral and EEG markers relevant for MI-based Brain–Computer Interfaces (BCIs). Sixteen healthy young adults performed a walking MI task in VR while embodied in either a young or elderly gender-matched avatar. Agerelated beliefs, embodiment, mental chronometry, and Alpha event-related desynchronization/synchronization (ERD/ERS) were assessed. Participants endorsed agerelated physical decline stereotypes and reported comparable embodiment across conditions. MI duration tended to be consistently longer when embodied in the elderly avatar. However, ERD/ERS analyses revealed modest and highly variable responses, with no consistent neural modulation across conditions.
• Deep learning based generalizable articulatory feature extraction for speech- BCI applicationsWu, Ruoling; Berezutskaya, Julia; Freudenburg, Zachary V.; Ramsey, Nick F.; 10.3217/978-3-99161-093-9-021Objective: Speech brain-computer interfaces (BCIs) often rely on paired neural-speech data, which is challenging to acquire from individuals with vocal tract paralysis. We aim to bridge this gap by proposing an across-subject framework that synthesizes intelligible speech using generalizable articulatory features extracted from an able-bodied cohort. Approach: We develop a subject-independent articulatory-to-speech model to extract generalizable articulatory features from real-time MRI (rtMRI) videos of a healthy cohort and use these features to synthesize text. Main Results: Our framework successfully synthesized text from the extracted generalizable articulatory features by achieving a Phoneme Error Rate (PER) of 18.9 % on unseen subjects. The results confirm that the extracted generalizable features remain robust and decodable across different subjects. Significance: By reducing the dependence on subject-specific articulatory data, our work offers a viable pathway for developing generalizable speech BCIs that could be used by those who can no longer articulate.
• Optimizing Bit-wise Linear Decoding for Imperceptible Code-Modulated Visual Evoked Potentials (I-c-VEP)Fodor, Milan A.; Tangermann, Michael; Thielen, Jordy; Volosyak, Ivan; 10.3217/978-3-99161-093-9-022Visual evoked potential (VEP)-based brain-computer interfaces (BCIs) are a well-established paradigm in the field, yet they remain limited by a fundamental usability constraint: they rely on perceptible low-frequency flicker, which induces visual fatigue and discomfort. To address this limitation, we previously proposed modulating an imperceptible high-frequency (60 Hz) carrier based on a pseudo-random binary code sequence. The resulting imperceptible code-modulated VEP (I-c-VEP) paradigm substantially reduces perceived flicker compared to conventional c-VEP approaches, while preserving key advantages over steady-state VEP (SSVEP) systems, namely efficient target scalability and robustness to narrowband interference. In this study, we explore whether simple, efficient linear decoding pipelines suffice for bit-wise I-c-VEP classification across 16 participants. By validating our pipelines on strictly independent, temporally held-out data, we demonstrated that approaches based on linear ridge regression and linear discriminant analysis can achieve over 97% mean trial accuracy. These findings provide a solid foundation for efficient and high-performance real-time implementations of the I-c-VEP paradigm.
• Ethical Considerations for User-Centric iBCI Studies: A Thematic ReviewThomas, Alex; Patel, Rishan; Podmore, Joshua James; Scherer, Reinhold; Carlson, Tom; 10.3217/978-3-99161-093-9-023Implanted brain-computer interfaces (iBCI) have been transformative for patients with paralysis, substantively restoring independent function in key domains. Despite this, the experimental nature and surgical complexity of implants means only 67 patients received long-term implants across published studies between 1998–2023. As the field grows, understanding what users actually want and need from these systems becomes critical to shaping future studies and applications. We performed a reflexive thematic review on the publicly available transcript of ’The Lived Experiences of BCI Users’ panel, conducted with three implanted BCI users, identifying 6 key themes: Purpose, Support, Understanding, Connection, Restriction, and Ethics. For each theme we discuss the core ideas with supporting quotes, derive best practice suggestions for future user-centred studies, and compare findings to existing literature, providing insight into the participant experience for researchers without direct exposure to user studies.
• EEG2EMG-Net: A Deep Learning Framework for Continuous EEG-to-EMG Estimation for Real-World ApplicationsMarcos-Martínez, Diego; Santamaría-Vázquez, Eduardo; Martínez-Cagigal, Víctor; Graz University of; Müller-Putz, Gernot; Hornero, Roberto; 10.3217/978-3-99161-093-9-024Estimating muscle activity from invasive neural recordings has been researched for decades with encouraging results. Such brain-computer interfaces (BCIs) are promising, as they offer an alternative pathway for motor control in patients with spinal cord injury. However, this application remains largely unexplored with non-invasive techniques, such as electroencephalography (EEG). Current non-invasive approaches typically rely on within-subject training, requiring both input (EEG) and output (electromyography, EMG) signals for calibration. This strategy is impractical for paralyzed patients unable to produce the required EMG. To address this, we developed EEG2EMG-Net a deep learning framework for continuous EEG-to-EMG estimation tailored for real-world clinical use. Our approach introduces a subject-adaptation strategy using synthetic EMG templates, eliminating the need for patient-specific muscle recordings during calibration. Results from 20 healthy participants show that the fine-tuning method allows accurate estimation of the EMG (pearson correlation coefficient: 0.69±0.32) while suppressing spurious activations during resting-state. These findings open new research avenues for non-invasive BCI alternatives aimed at assisting motor function in individuals with severe impairments.
• Neural and Clinical Variability in Immersive VR–BCI Training: A Case-Series Study with Stroke PatientsValente, Madalena; Oliveira, Ines; Fernandes, Jean-Claude; Almeida, Ana Isabel ; Figueiredo, Patricia; Vourvopoulos, Athanasios; 10.3217/978-3-99161-093-9-025Immersive virtual reality–based brain–computer interface (VR–BCI) systems have emerged as promising tools for post-stroke motor rehabilitation, providing embodied feedback and taskspecific training. However, reliable neurophysiological predictors of clinical responsiveness remain unclear. This case-series study investigated whether longitudinal modulation of event-related desynchronization (ERD) during VR–BCI training predicts upper-limb motor improvement in individuals with sub-acute and chronic stroke. Five participants completed 12 VR–BCI sessions over four weeks, with clinical assessments conducted pre- and post-intervention. ERD was extracted using individualized frequency bands, and linear mixed-effects models were applied to estimate subject-specific ERD trajectories, including baseline magnitude and sessionto- session change. These features were then examined in relation to motor recovery. Robust Alpha ERD was observed across participants, indicating consistent sensorimotor engagement. However, ERD progression across sessions did not significantly predict clinical gains, consistent with prior findings. Finally, BCI accuracy also showed substantial inter-subject variability and was not directly associated with motor improvement.
• Learning brain connectivity features for improving BCI performancePavaux, Marion; de Vico Fallani, Fabrizio; La Rocca, Daria; 10.3217/978-3-99161-093-9-026Brain computer interfaces (BCIs) rely on the decoding of information embedded in brain signals for applications ranging from neurofeedback to device control. While power spectral features are widely used in motor imagery-based BCIs, recent evidence suggests that considering network interactions between sensors could improve performance. In practice however, connectivity between different brain signals is impaired by the high variability of classical estimators for short noisy segments. To address this issue, we propose a deep learning framework trained to recover ground-truth imaginary coherence from complex simulated signals. Here, we show that our framework learns to reconstruct smooth coherence spectra, designed as Gaussian mixtures. Evaluation on an electroencephalography dataset across three motor execution and imagery tasks, shows that connectivity matrices derived from the proposed method consistently outperform those obtained withWelch’s estimator. These results demonstrate the feasibility of using deep learning to obtain stable informative coherence estimations in sparse and noisy contexts, supporting the use of functional brain networks for BCI applications.
• A Systematic Overview of EEG-Detectable Control-Related Mental States and Processes in HCIMatthias, Eidel; Settgast, Tomko; Silveira, Sarita Teja; Krol, Laurens R.; Wagner, Johanna; Kübler, Andrea; 10.3217/978-3-99161-093-9-027We report our endeavor to systematically map and describe all EEG-detectable mental states and processes (MS) with a focus on Human-Computer Interaction (HCI) from existing literature. Methods: Literature was collected in two steps. First, we performed a large-scale query against three relevant literature databases to systematically retrieve all articles dealing with EEG-detectable MS within a HCI context. Second, we broadened our scope to encompass all MS without the restriction to HCI. We obtained and screened the full list of paradigms from the #EEGManyLabs consortium and performed a second database query, looking specifically for influential reviews and metaanalyses about EEG-detectable MS. Results: We provide a limited overview, focusing on mental states relating to control processes, based on 1024 screened articles, thus far. Specifically, we describe the different MS within this category and their respective elicitation paradigms and neural correlates. Significance: While originally created for the NAFAS project, this overview may serve as a reference for anyone working with EEG-detectable mental states.
• Finger abduction trajectory prediction from high-density ECoGFaes, Axel; Merino, Eva Calvo; Mirsaeedi, Mani; Van Hoylandt, Anaïs; Keirse, Elina; Theys,Tom; Van Hulle, Marc M.; 10.3217/978-3-99161-093-9-028A case study is conducted to demonstrate that not only finger flexion but also finger abduction trajectories can be decoded from high density electrocorticography (ECoG) recordings. Two patients, temporarily implanted with high-density ECoG grids as part of their clinical workup, performed three cued tasks: single finger flexion, finger abduction, and sign language alphabet gestures. The extended Block-Term Tensor Regression (eBTTR) model is trained on simultaneous ECoG and data glove recordings to predict continuous finger trajectories. Performance on the flexion task serves as a baseline for comparison with abduction decoding. The sign language task involves realistic, complex finger movements combining both flexion and abduction. We show that finger abduction trajectories can be decoded from ECoG with precision comparable to single finger flexion, even during sign language gestures that involve joint finger flexions and abductions. To our knowledge, this is the first demonstration that finger abduction trajectories can be decoded from ECoG.
• EEG-Based Detection of Continuous and Discrete Hand Movements for Cursor ControlCrell, Markus R.; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-029Continuous cursor control for non-invasive brain-computer interfaces (BCIs) has recently received widespread attention. However, the usability of such systems for patients with e.g., neuromuscular diseases remains limited while a selection mechanism is not simultaneously provided. We therefore attempt to asynchronously detect and discriminate between continuous and discrete hand movements to enable intuitive cursor control and click possibilities for cursor BCIs. Our results show that discrimination between discrete and continuous motions is possible with high accuracy and a low rate of false positive detections. We propose that the introduced method can be applied to non-invasive cursor control BCIs to add a click functionality to the cursor movement.
• Riemannian approaches for ECoG-based Motor Imagery BCI decoding using spatial covariance matrices : a comparative studyEtienne, Mathis; Martel, Felix; Bonnet, Stéphane; Aksenova, Tetiana; 10.3217/978-3-99161-093-9-030Symmetric positive definite (SPD) covariance matrices have demonstrated strong potential for motor imagery (MI) decoding in EEG-based brain-computer interfaces (BCIs), yet their application to electrocorticography (ECoG) data remains largely unexplored. In this study, we evaluate and compare multiple classification approaches leveraging the Riemannian geometry of SPD matrices for ECoG-based MI decoding, across two paradigms: upper limb (3-class) and lower limb (3- class) motor imagery. Spatial covariance matrices were estimated from signals filtered in two task-specific frequency bands and used as input features. Three families of classifiers were compared: tangent space mapping (TSM) combined with machine learning models (LR, LDA, RSW-NPLS, LightGBM) and a multilayer perceptron (MLP); classifiers operating directly on the Riemannian manifold (MDM, FgMDM); and an SPD-aware neural network (SPDNetBN). Models were trained and tested offline on data from two implanted patients using lagcorrected balanced accuracy as the performance metric. FgMDM and TSM+RSWPLS achieved the highest performance for the upper limb paradigm (up to 91.6% and 89.5%, respectively), while SPDNetBN demonstrated superior accuracy for the lower limb paradigm (up to 69.4%).
• Why Simple Models Still Matter: Adaptive sLDA for Non-Stationary EEG DecodingEgger, Johanna; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-031Electroencephalography (EEG)-based brain-computer interfaces (BCIs) are strongly affected by non-stationarities, causing covariate shifts between calibration and deployment and leading to performance degradation over time. While deep learning approaches address this issue, they typically require large training datasets that are often unavailable in practical BCI scenarios. We therefore investigate lightweight online adaptation strategies for shrinkage linear discriminant analysis (sLDA), a simple yet robust classifier for lowdata EEG settings. We evaluate three unsupervised methods: exponential moving average feature normalization (FNorm), pooled mean adaptation of the classifier bias (PMean), and pooled mean with covariance adaptation (PMean+Cov). Across six sessions from 20 participants, without FP/min control, true-positive rates rose by 14.27% (FNorm), 29.13% (PMean) and 5.87% (PMean+Cov), while falsepositives per minute changed to 3.71 (+2.19) for FNorm, 11.95 (+10.44) for PMean, and 1.05 (-0.47) for PMean+Cov (not statistically significant). PMean+Cov achieved the best balance between sensitivity and false positives, making it most suitable for real-time BCI speller applications.
• Wrapped One-Class Riemannian EEG Classifier for BCI-Detection of Anesthetic StatesCueva, Valérie Marissens; de Surrel, Thibault; Laurent Bougrain; Bidgoli, Seyed Javad; Cheron, Guy; Alvarez, Ana Maria Cebolla; Meistelman, Claude; Lotte, Fabien; Yger, Florian; Rimbert, Sébastien; 10.3217/978-3-99161-093-9-032Brain-computer interfaces based on ElectroEncephaloGraphy (EEG) often represent efficiently brain signals as covariance matrices and leverage Riemannian geometry for classification of mental states. However, current approaches typically require multiple classes for training. In many applications such as intraoperative monitoring of consciousness during anesthesia, only data from one class, such as the awake state, may be available, requiring one-class classification methods. We propose a novel Riemannian one-class classifier called the One-Class Wrapped Gaussian. Unlike existing methods that rely solely on the Riemannian mean, our approach incorporates second-order statistical information by using an anisotropic Gaussian-like distribution on the manifold of covariance matrices. We validated our method on EEG data from 19 patients undergoing general anesthesia. Results show that our classifier significantly outperforms state-of-the-art one-class methods for distinguishing between awake and anesthetized states, for clinically relevant electrode numbers. We further demonstrate the robustness of our approach by testing configurations with reduced electrode numbers, confirming its feasibility for real-world surgical settings where electrode placement is constrained.
• Filling the gap: sex-specific differences in post-task interval dynamicsLeitner, Michael; Wriessnegger, Selina C.; 10.3217/978-3-99161-093-9-033Electroencephalography (EEG) is widely used to investigate neural correlates of cognitive work load, fatigue, and stress through changes in oscillatory band power. While many studies have focused on neu ral activity during task execution, the post-task resting period following cognitive effort remains comparatively underexplored, despite its potential relevance for under standing how cognitive resources are restored. More over, little is known about whether neural dynamics dif fer between male and female participants. In this study, we analyzed EEG band power during the post-task rest ing interval following a cognitively demanding working memoryparadigm. Relativebandpower(rBP)wasexam ined across multiple frequency bands, scalp regions, and rest epochs to characterize temporal and spatial patterns. Significant sex-related differences were observed in the theta, alpha, and beta bands. Females exhibited higher theta activity, particularly in central regions, whereas males showed elevated alpha band power. Across rest ing epochs, alpha activity displayed the most pronounced increase, suggesting progressive reallocation of atten tional resources. Topographic analyses further revealed region-specific differences in central, parietal, and occip ital areas. These findings suggest that post-task interval dynamics exhibit distinct oscillatory characteristics that vary between sexes. Future studies might consider such baseline phases leading to improved EEG-based detection of cognitive states and inform the design of neuro adaptive systems.
• Is User-Independent MI-BCI Performance Influenced by Gender Proportion in Training Users?Kojima, Simon; Lotte, Fabien; 10.3217/978-3-99161-093-9-034In motor-imagery brain–computer inter faces (MI-BCI) research, how user gender influences BCI classification performance and ERD/S responses is not yet fully understood. Thus, investigating the impact of gender on BCI systems is an important step toward real izing BCIs that are accessible to all users. In this study, we examined how varying the proportion of female users in the training data in cross-user MI-BCI affects BCI clas sification performance. The results (on 139 users from 2 data bases) revealed positive correlations (0.33–0.71) be tween the proportion of female users in the training data and BCI performance, and showed that including more female users in the training data was associated with per formance improvements of up to 5.5%. These findings indicate that machine-learning models in MI-BCI may be influenced by gender-related biases and highlight the im portance of considering gender distribution in the design of future BCI research.
• Non-invasive measurement of somatosensory evoked spinal cord potentialsGolser, Bernhard; Oberndorfer, Markus; Kostoglou, Kyriaki; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-035The spinal cord is more than a mere re lay station between the brain and the peripheral nervous system. Recent research indicates that complex process ing already occurs at the spinal level. The non-invasive recording of spinal cord activity acquired with high density surface electrode arrays may advance our under standing of spinal network dynamics and the integration of spinal data could enhance existing brain–computer in terfaces. In the present study, electroencephalography and electrospinography were used to measure cortical and spinal somatosensory evoked potentials elicited by median and tibial nerve stimulation in 10 healthy indi viduals. In contrast to previous studies, participants were seated upright with no external support of the head and stimulus intensities were kept strictly below the move ment threshold. Under these more physiologically real istic settings, two of the three well-characterized spinal cord components (N13 and P22) were successfully de tected, indicating a broader variety of potentially success ful experimental setups.
• Towards thermal imagery-based brain-computer interfacesLefeuvre, Théo; Savalle, Emile; Petit, Jimmy; Macé, Marc J-M.; Lécuyer, Anatole; Pillette, Léa; 10.3217/978-3-99161-093-9-036Mental task-based brain-computer inter faces (BCIs) enable users to control a system by volun tarily modulating their brain activity without overt move ment. Research in the field has focused mainly on motor imagery, whereas tactile imagery, defined as the imagi nation of tactile stimuli on the skin, remains compara tively underexplored. Although pressure and vibratory imagery have received growing attention, other modal ities such as thermal imagery remain largely neglected, despite their potential clinical relevance. In this EEG based BCI study, healthy participants learned to con trol a BCI by imagining a warm sensation in the right hand, representing, to our knowledge, the first thermal imagery-based BCI user training. To facilitate mental representation, participants received visual, thermal, or combined visuothermal instructions. Thermal imagery elicited event-related desynchronization, primarily over left sensorimotor regions, and online performance im proved significantly across runs. Comparisons between instruction modalities suggested trends favoring visual and thermal guidance. Together, these findings support the feasibility of thermal imagery as a promising alterna tive control paradigm for BCI applications.
• Movement-Onset-Aligned EEG for Generalizable Arm and Hand Movement Decoding: A Large-Scale Multi-Experiment DatasetSterk, Kathrin; Kostoglou, Kyriaki; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-037This work presents a curated, movement onset-aligned electroencephalography (EEG) dataset comprising nine experiments with 188 able-bodied par ticipants performing hand and arm movement execution tasks as well as rest conditions. Movement onsets were aligned using auxiliary sensors. All recordings were uni formly processed using causal filtering to ensure com patibility with online brain-computer interface (BCI) ap plications. Further processing steps include resampling, independent component analysis (ICA) for removal of ocular and noise components, and strict epoch rejec tion. Across experiments, the datasets include recordings with different channel configurations, covering 94 dis tinct EEG channels. In total, the dataset comprises over 95,000 annotated trials with detailed movement labels and is provided in a standardized HDF5 structure to fa cilitate flexible data selection in machine learning work flows. Signal analyses confirm the presence of physiolog ically plausible movement-related patterns over sensori motor regions. This resource aims to support the devel opment of cross-participant and cross-experiment gener alizing movement decoders and transfer learning in EEG based BCIs. The data will be released in the near future.
• Benchmarking Machine Learning strategies to classify Steady-State Auditory Evoked Potentials.da Silva, Henrique Lefundes; Guého, Lenaïg; Plapous, Cyril; Bougrain, Laurent; Hénaff, Patrick; Nicol, Rozenn; 10.3217/978-3-99161-093-9-038This study benchmarks classification pipelines for auditory Brain–Computer Interfaces (BCIs), with the objective of detecting the modulation frequency of amplitude-modulated sinusoidal auditory stimuli in electroencephalographic recordings. For this purpose, state-of-the-art Steady-State Visually Evoked Potential (SSVEP) classification algorithms are adapted to the au ditory domain. This choice is motivated by the neuro physiological similarities between SSVEP and Steady State Auditory Evoked Potential (SSAEP), particularly their shared mechanism of neural entrainment to peri odic stimuli, despite differences in stimulation frequen cies and the cortical regions involved. In this context, both preprocessing strategies and classification models are systematically assessed to identify optimal parame ter configurations. Among the filtering techniques exam ined, a narrow-band Bessel filter yielded the highest ac curacies, while the effect of temporal windowing was less pronounced under narrow-band conditions but remained beneficial for broader bandwidths. Riemannian-based methods, particularly tangent-space classifiers, consis tently outperformed alternative approaches. Neverthe less, cross-subject classification proved more challeng ing due to pronounced inter-subject variability. These findings provide practical recommendations for auditory BCIs.
• Capacity-normalized Shannon graph complexity as an EEG biomarker for event-level cognitive load estimation in adaptive BCIsPascual-Roa, Beatriz; Santamaría-Vázquez, Eduardo; Marcos-Martínez, Diego; Ruíz-Gálvez, C. Rubén; Martínez-Cagigal, Víctor; Hornero, Roberto; 10.3217/978-3-99161-093-9-039Cognitive load (CL) reflects the mental resources required to perform a task. Adaptive brain computer interfaces (BCIs) could benefit from EEG biomarkers that estimate CL with sufficient temporal granularity to support online adaptation. However, many EEG-based CL studies still define CL according to pre defined task levels or objective task difficulty, which may not impose comparable cognitive demands across users. This assumption is especially problematic when individ uals differ in working-memory capacity. We propose an event-level and capacity-normalized framework based on EEG functional connectivity in frontal and parietal re gions. During the Corsi Block-Tapping Test, each en coded item was labeled according to its position rela tive to each participant’s individual maximum working memory capacity, allowing cognitive demand to be ex pressed as a percentage of individual capacity. Functional connectivity was quantified using orthogonalized am plitude envelope correlation (AEC) across theta, alpha, low- and high-beta, and gamma frequency bands, while network organization was summarized through Shannon graph complexity (SC). The approach was evaluated in 62 cognitively healthy older adults (mean age = 70.4±2.9 years). Across all frequency bands, SC showed a pro gressive increase as task demands approached individ ual working-memory capacity. Capacity-normalized CL stages differed significantly after FDR correction (p < 0.05), with mostly large effect sizes (r >0.5). These find ings suggest that SC, combined with capacity-normalized event-level labeling, may provide a compact and online compatible EEG biomarker for CL-aware BCI adaptation in cognitively healthy older adults.
• Interpretability of CNN-Based EEG Decoding During Mirror Box Training: A Pilot Study Using CWT Spectrograms and Explanaible AIHodúr, Adrián; Rosipal, Roman; Evetović, Nina; Rošťáková, Zuzana; Jakubík, Jozef; 10.3217/978-3-99161-093-9-040Neural networks are increasingly applied in brain-computer interface research; however, their decision-making process is often considered difficult to interpret. This pilot study investigates whether a con volutional neural network can provide not only accu rate but also neurophysiologically meaningful classifi cation of electroencephalogram signals recorded during mirror box training. Two conditions, rest and left hand movement, were analyzed with a focus on interpretabil ity. EEG signals from contralateral electrode C3 were transformed into time-frequency representations using the Continuous Wavelet Transform, and the resulting spectrograms were used as input to the convolutional neu ral network (CNN). To interpret the model’s decisions, Gradient-weighted Class Activation Mapping and other methods were applied, enabling identification of discrim inative time-frequency regions. By relating these pat terns to established neuroscientific knowledge, we assess whether the model relies on physiologically meaning ful features, providing insight into brain dynamics dur ing mirror-induced motor activity and supporting inter pretable BCI applications.
• Fast-Calibrated c-VEP BCI with Comfort-Optimized Stimulation for Robust Control on Screen and Immersive VR Using Semi-Dry EEGDehais, Frédéric; Maggi, Emilio; Gomel, Jules; Tresols, Juan Torre; Cimarosto, Pietro; Velut, Sébastien; Cabrera-Castillos, Kalou; 10.3217/978-3-99161-093-9-041Code-modulated visual evoked potentials (c-VEPs) are fast and reliable control signals for reactive brain-computer interfaces (rBCIs), but real world use is of-ten constrained by visually intrusive flicker and the need for wet EEG. We present a usability-oriented c-VEP rBCI that combines visually comfortable, low-salience textured stimulation, low-rate coding with asynchronous evidence accumulation to allowqww flexible decision times, and a semi-dry EEG setup to improve wearability while maintaining signal quality. We evaluated the sys-tem in two settings: a PC condition consisting of an 11-class keypad typing task on a standard computer screen, and a VR condition consisting of a simple shell game. Despite using only 33 s of calibration data, we achieved 100% accuracy in both conditions. Mean decoding time was approximately 4.3 s in the PC condition and 5.4 s in the VR condition, consistent with additional head mo tion and rendering constraints in VR. Participants also re ported high visual comfort, low visual fatigue, and min imal peripheral distraction from the flicker, supporting the usability of the proposed stimulation. Overall, these results demonstrate fast-calibrated, robust, and visually comfortable c-VEP control that generalizes from conven tional displays to immersive VR with a practical semi-dry EEG configuration.
• Towards Reliable Out-of-Lab pBCIs: Riemannian Geometry Framework for Mitigating Covariate Shift in Real-World Passive Brain-Computer InterfacesGermano, Daniele; Ronca, Vincenzo; Capotorto, Rossella; Ercole, Simone; Borghini, Gianluca; Di Flumeri, Gianluca; Aricò, Pietro; 10.3217/978-3-99161-093-9-042Passive brain-computer interfaces (pBCIs) are promising tools for real-world, out-of-lab monitoring of cognitive states (e.g. mental workload). However, their reliability (e.g. classification accuracy) is often compromised by covariate shift caused by changes in headset positioning among different sessions. This study introduces a Riemannian-based correction method de signed to mitigate the negative effect of the shift. We evaluated the framework by training a random forest classifier on a source domain to classify two mental workload levels and then tested on a target domain (rep resenting a channels shift) both before and after applying the transformation. The validation involved three levels of analysis: (i) classification performance via area under curve (AUC) analysis, (ii) statistical assessment of man ifold alignment through Riemannian distance, and (iii) correlation between geometric correction and AUC re covery. The application of the correction significantly improved workload classification, significantly over coming AUC scores compared to the uncorrected target data. Statistical analysis revealed a significant reduction in Riemannian distance between the source and target manifolds after correction (𝒑 = 𝟐.𝟑𝟔𝒆−𝟓). Furthermore, correlation analysis showed a strong positive relationship between geometric alignment and performance gain (Pearson correlation 𝑹 = 𝟎.𝟔𝟐𝟖).
• Decoding Spoken Syllables from EEG Data via Contrastive Embedding AlignmentCortecchia, Tommaso; Volpato, Pietro; Tortora, Stefano; Menegatti, Emanuele; Tonin, Luca; 10.3217/978-3-99161-093-9-043We explore the potential of electroen cephalography (EEG)-based brain-computer interfaces as an alternative communication system for people with speech impairments. Instead of considering single words or phonemes, we exploit the syllabic structure of the Italian language as the fundamental building block of speech. We collected data from 21 participants as they pronounced and syllabified a set of sentences. We then trained a neural network to learn a latent representation of EEG segments through contrastive alignment with em beddings extracted from a pre-trained automatic speech recognition model. We framed the evaluation as an infor mation retrieval task, utilizing a top-k retrieval within the EEG latent space, comparing a global model to subject specific models. We achieved an average Top-10 accu racy of 37.4% for the multi-subject model and of 38.2% for subject-specific models across a 192-class vocabulary. The proposed framework provides a scalable approach to non-invasive speech decoding, enabling greater expres siveness and speed compared to traditional solutions in assistive communication.
• Mirror-Box Pretraining to Evoke Sensorimotor Activation to Facilitate Motor Imagery Detection in Brain-Computer Interface Stroke RehabilitationEvetovic, Nina; Rosipal, Roman; Hodúr, Adrián; Rošťáková, Zuzana; 10.3217/978-3-99161-093-9-044Calibration of motor imagery (MI) brain computer interface (BCI) systems in stroke patients is challenging due to limited voluntary movement and vari able imagery ability. We propose mirror-box (MB) train ing as a physiologically grounded pretraining strategy to induce contralateral sensorimotor activation and extract subject-specific oscillatory components for MI detection. Electroencephalography (EEG) was recorded during six MB sessions in a chronic post-stroke participant. Parallel Factor Analysis (PARAFAC) was applied to decompose multi-channel EEG into spatial, temporal, and spectral components. Trial-pair normalization relative to the cor responding rest segment was used to enhance detection of task-related modulation. Distinct µ and β sensorimotor rhythms were identified, showing significant desynchro nization during observation and movement compared to rest. An individualized linear discriminant analysis clas sifier achieved 98.8% ±0.6% accuracy in discriminat ing rest from motor engagement. These findings sup port MB pretraining as a robust strategy for physiologi cally informed MI-BCI calibration and future BCI Head Mounted Display rehabilitation applications.
• Unprotected Minds: A Clinical Neuropsychology Perspective on the Ethics of Emerging NeurotechnologiesLupașcu, Alberta Maria; Flórez Rojas, Maria Lorena; Enriquez-Geppert, Stefanie; 10.3217/978-3-99161-093-9-045The rapid expansion of EEG-based neurotechnology from clinical and research settings into consumer and workplace environments has created new ethical challenges regarding the interpretation and governance of neural data. While clinical applications operate under strict therapeutic intent and expert oversight, and research is similarly governed by rigorous ethical and regulatory frameworks, non-clinical uses often rely on simplified metrics, unclear consent processes, and inadequate safeguards. This review goes beyond a narrative approach by examining the ethical gaps that arise when EEG-based technologies operate across divergent contexts. Through a biomedical ethics framework, the analysis reveals a critical failure to protect both neural signals and their interpretation once these technologies move outside the clinic. Central to this issue is the critical role of the clinical neuropsychologists, whose expertise enables contextualizing the neural signals while preventing the oversimplification of EEG data, and the overgeneralization of probabilistic brain markers. The findings underscore implications for three key stakeholder groups: neuropsychologists, who can establish interpretative standards; the BCI community, which is critical for ensuring the scientific validity and transparency of EEG-inferred cognitive states; and policymakers, who must address the regulatory gaps by extending protections beyond neural data collection.
• Circadian alignment modulates mental tasks BCI performanceDreyer, Pauline; Kojima, Simon; Bourdil, Manon; Bechon, Loic; Roy, Raphaëlle N.; Lotte, Fabien; 10.3217/978-3-99161-093-9-046Active brain–computer interfaces (aBCIs) exhibit substantial performance variability both across and within participants. While many sources of vari ability have been investigated, circadian influences re main largely unexplored in aBCI control. We investi gated whether time of day and chronotype (individual’s circadian preference for sleep and activity timing across the day) jointly modulate BCI performance (classifica tion accuracy). Twelve participants completed 6 EEG based BCI sessions each alternating between morning and afternoon sessions, performing 3 mental tasks (mo tor imagery, mental calculation, letter/word association). Online performance was analyzed using generalized lin ear mixed models with binomial distribution and random intercepts for participants and sessions. Overall perfor mance was above chance level, with a significant de crease across runs but no linear trend across sessions. A significant interaction between time of day and chrono type was observed: morning-type participants tended to perform better in the morning, whereas evening-type par ticipants showed relatively higher performance in the af ternoon, consistent with the circadian alignment.
• Introducing simulation-based inference to c-VEP BCI: A feasibility studyVelut, Sébastien; Corsi, Marie-Constance; Chevallier, Sylvain; Dehais, Frédéric; Thielen, Jordy; 10.3217/978-3-99161-093-9-047Code-modulated visual evoked-potential (c-VEP) based reactive brain-computer interfaces (BCIs) deliver high information-transfer rates with minimal cal ibration, yet performance collapses when generalizing across participants. Additionally, only few studies fo cus on understanding the neural mechanisms underlying c-VEP, which remain poorly understood. In this paper, we introduce the simulation based inference (SBI) frame work to the BCIfield. SBIallowstooffloadcalibration on simulated synthetic data generated with parameters that may represent a large cohort of different participants. We assess SBI’s feasibility to provide competitive classifica tion performance on empirical data as well as inferring response parameters, while being calibrated only on syn thetic data. We achieved 82%classification accuracy with a within participant setup using a c-VEP based BCI, and a mean squared error of 0.11 between the empirical visual evoked potential and the inferred one. These results sug gest the feasibility of SBI for BCI, and could deepen our understanding of the mechanisms underlying the c-VEP. We present the current limitations of the SBI framework and give future directions to address these.
• BCI-based Intelligent Tutoring Systems: A Literature Review and Meta-Analysis on Neuroadaptive ITSPetit, Jimmy; Macé, Marc J-M.; Lécuyer, Anatole; Pillette, Léa; 10.3217/978-3-99161-093-9-048Intelligent tutoring systems (ITS) are arti ficial systems that provide personalised instruction, re lying on student models to monitor cognitive states, such as mental workload. Recent studies have increasingly en hanced this monitoring through the use of passive brain computer interfaces (BCI). In this review, we focused on ITS that use BCI metrics to adapt their interven tions in real time. Following PRISMA 2020 guidelines we searched PubMed, Web of Science, Scopus, Ope nAlex, and Semantic Scholar databases. We conducted a risk-of-bias assessment using RoB 2.0, and performed a meta-analysis to evaluate the impact of BCI-based ITS on learning outcomes. We found a moderate-to-large pooled effect size (SMD = 0.79) with a 95% confidence interval [−0.08;1.65] suggesting a potential benefit of BCI-based ITS. However, the wide confidence interval that includes zero indicates substantial uncertainty. As this field is still emerging, with most studies published after 2015, further research is needed to strengthen the evidence base and support more definitive conclusions.
• A controlled evaluation of dry textile EEG electrodes using simulated P300 and SSVEP recordings with a gelatin-based skin phantomLeung, Jason; Qureshi, Faiz Uddin; Chau, Tom; 10.3217/978-3-99161-093-9-049Dry textile electrodes offer reduced setup time and improved user comfort, making them attractive for everyday electroencephalography (EEG) recordings. However, dry electrodes typically exhibit higher electrode-skin impedance, so their signal quality must be carefully evaluated for brain-computer interface (BCI) applications. In this study, dry textile EEG electrodes were evaluated using a gelatin-based skin phantom. Simulated EEG recordings during P300 and steady-state visual evoked potential (SSVEP) tasks were delivered through electrodes embedded in the phantom model. Signals recorded from textile electrodes placed on the surface of the phantom show strong agreement with those from Ag-AgCl electrodes. P300 components were clearly observable in the event-related potentials, and distinct spectral peaks were detected at the stimulation frequencies in the SSVEP recordings. These results suggest that dry textile electrodes provide signal quality comparable to conventional Ag-AgCl electrodes under controlled conditions, warranting further investigation into their feasibility for BCI applications.
• Movement-Related Interhemispheric Connectivity Dynamics and their Link to Asynchronous EEG-BCI PerformanceKostoglou, Kyriaki; Crell, Markus R.; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-050Directed interactions between bilateral sensorimotor cortices are important for voluntary movement, yet their relevance for asynchronous brain computer interface (BCI) control remains unclear. Using EEG, we quantified time–frequency directional connectivity between sensorimotor regions during cue based and self-paced hand movements in 25 participants using time-varying multivariate autoregressive models. Connectivity from the contralateral to the ipsilateral cortex was consistently stronger and positively associated with movement detection performance. Connectivity increased in low-frequency bands around movement onset and in the beta band after movement execution. These findings suggest that interhemispheric EEG connectivity reflects motor network coordination and may serve as a neurophysiological marker of successful asynchronous BCI control.
• Tracking Vigilance Fluctuations with Aperiodic Semi-Transparent Visual FlickersCimarosto, Pietro; Dehais, Frederic; 10.3217/978-3-99161-093-9-051We investigated whether aperiodic semi transparent textured rapid visual stimuli (RVS) can pro vide an unobtrusive neural probe of vigilance fluctua tions during a 60-min Mackworth clock vigilance task. Six healthy participants performed the task while EEG was recorded continuously and nearly imperceptible vi sual stimuli were superimposed on the interface. We analyzed event-related potentials (ERPs) and conven tional band-power features at both the group and single trial levels. Preliminary results showed significant hit versus-miss differences in the N200 and P300 amplitudes elicited by the stimuli, whereas classical frontal theta and parieto-occipital alpha features did not significantly distinguish the two conditions. Single-trial classifica tion yielded modest performance overall, although ERP based features appeared more informative than conven tional band-power markers, and combining evoked and spectral features slightly improved hit-versus-miss dis crimination. These preliminary findings suggest that ape riodic semi-transparent RVS may offer a promising and unobtrusive approach for probing vigilance fluctuations.
• Topological Higher-Order Features for Brain-Computer InterfacesSkhiri, Wafa; de Vico Fallani, Fabrizio; 10.3217/978-3-99161-093-9-052Integrating features from the way different brain signals mutually interact has been recently used to understand and improve brain-computer interfaces (BCIs). How-ever, most existing approaches have focused on pairwise network features, which neglect possibly existing inter-actions involving more than two brain areas. Increas-ing evidence suggests that higher order (HO) interactions arising from the simultaneous coordination of multiple groups of regions, bring complementary information discrimination. for brain states Here, we investigated whether higher-order brain net work features extracted from EEG signals can improve the classification of simple motor imagery-based BCI tasks. To this end, we developed a method that combines spectral graph embedding and topological data analy sis to extract HO structures from functional connectivity networks estimated using imaginary coherence between EEG signals. We evaluated the ability of these features to discriminate motor imagery (MI) and resting-state EEG activity in a typical 1D cursor task control. Results showed that MI significantly increases the involvement of sensorimotor electrodes in HO triadic schemes as compared to resting states. These changes were particularly present in beta band frequency ranges. Furthermore, we found that in 16 out of 19 participants, such HO features achieved higher accuracy than both local power spectra and connectivity. These findings suggest that topological characterizations of functional brain networks capture behaviorally rele vant aspects of neural organization and offer a promising direction for incorporating higher-order interactions into EEG-based BCI applications.
• ERP-XTTN: Interpretable Cross-Subject Error-Related Potential Classification via Cross-Attention to Data-Driven ERP PrototypesWyman, Charlotte Genevier; Hirshfield, Leanne; 10.3217/978-3-99161-093-9-053Error-related potential (ErrP) classifica tion methods for brain-computer interface deployment have separately addressed causal filtering, calibration free generalization, and interpretability, but no existing approach combines all three. We propose ERP-XTTN, a cross-attention architecture using fixed difference-wave prototypes with temporal windows derived from training data. A peak-detection algorithm identifies ErrP compo nents from the grand-average difference wave and sets window boundaries at smoothed zero-crossings. The pro totypes serve as keys in a cross-attention mechanism with no value projection, so classification depends entirely on interpretable attention routing. We evaluate via leave one-subject-out (LOSO) cross-validation on two datasets (BNCI Horizon 2020, 6 subjects; HRI Cursor, 11 sub jects) with causal filtering and no hyperparameter tuning. ERP-XTTN achieves 0.782/0.836 AUROC (BNCI/HRI) using three midline channels and remains competitive with two. Attention maps reveal class-discriminative routing aligned with event-related potential (ERP) com ponent timing on both datasets. These results show that data-driven prototypes can approach black-box perfor mance while preserving interpretability grounded in ERP component timing, without sacrificing causal filtering or calibration-free generalization.
• TREAT: a toolbox for trial selection and alignment of unlabeled neural dataGeukes, Simon H.; Bekius, Annike; Fabris, Federica; Freudenburg, Zachary V.; Ramsey, Nick F.; Branco, Mariana Pedroso; 10.3217/978-3-99161-093-9-054Brain–computer interfaces (BCIs) enable paralyzed individuals to control external devices using brain signals. For training decoders, trial-to-trial variability in response time may obscure meaningful task‑related neural activity. We introduce the TRial sElection and AlignmenT (TREAT) toolbox to provide an automated neural event onset detection method that supports both trial selection and event alignment, based on the slope of the neural response (neural slope marker, NSM). We validated TREAT using high‑density electrocorticography recordings from three able‑bodied individuals performing a speech task. We extracted highfrequency and alpha band power and compared the NSM to the voice onset times. For two of the three participants, the neural slope markers preceded voice onset by 0.4 seconds. Trial selection improved the precision of neural onset estimates across trials as the spread of NSM timing compared to voice onset time decreased. TREAT offers event-detection in the absence of behavioral markers and may thereby improve interpretation of neural activity and training of BCI decoding models.
• Selection and Validation of Spectral EEG Markers for Real-Time Monitoring of Attention and Cognitive Engagement in Learning Contextsde Carvalho, Pierre-Baptiste Mathieu; Corsi, Marie-Constance; Bougrain, Laurent; 10.3217/978-3-99161-093-9-055Electroencephalography (EEG) provides precise neural measures for brain-computer interface (BCI) applications monitoring cognitive states. This empirical study evaluates the sensitivity of established EEG spectral markers to distinguish sustained attention from cognitive engagement. Fifteen participants completed a sustained attention task (D2-R), a N-Back working memory task (2-Back), and an ecological video learning task. Results indicate that the sustained attention task induced significant modulations in the β/(α +θ) ratio, posterior theta, and occipital beta power. Conversely, the cognitive engagement task yielded no significant spectral deviations, highlighting potential overlaps in cognitive networks and the critical impact of task-order effects on post-effort baselines. Furthermore, linear mixed models revealed distinct, non-linear temporal dynamics during the ecological task, demonstrating that aggregating EEG metrics over the entire task duration obscures critical real-time state fluctuations. This study provides a rigorous framework for selecting appropriate EEG markers, emphasizing the necessity of continuous monitoring and controlled experimental designs in educational BCI applications.
• Beyond Visual Feedback: Isolating Vestibular Error-Related Potentials During Wheelchair OperationsVirgolini, Francesca; Simonetto, Piero; Tortora, Stefano; Menegatti, Emanuele; Tonin, Luca; 10.3217/978-3-99161-093-9-056The Error-Related Potential (ErrP) is a well-established neurophysiological marker of error perception, historically elicited through visual feedback in Brain-Machine Interfaces (BMIs). However, to our knowledge, no study has yet investigated whether such a response can be elicited in the complete absence of visual input, relying solely on vestibular feedback generated by physical movement. This study addresses this question by recording electroencephalographic (EEG) signals from participants during a powered wheelchair motion task, under two conditions: one with visual feedback available, and one with visual occlusion imposed through blindfolding. Our findings demonstrate that a recognizable ErrP waveform is elicited under full visual occlusion, exhibiting a morphology consistent with that observed in the sighted condition. Observed differences in latency and amplitude between conditions are discussed in relation to their experimental contexts. These results provide preliminary evidence that vestibular feedback alone constitutes a sufficient substrate for ErrP generation, opening a promising avenue for BMI systems not relying on visual input for implicit error monitoring.
• Expertise-Dependent Effects of Neurofeedback Design During Motor ImageryPachoud, Mathieu; Afankous, Mehdi; Segas, Effie; Izac, Margaux; N'Kaoua, Bernard; Jeunet, Camille; 10.3217/978-3-99161-093-9-057Motor imagery (MI) combined with neurofeedback (NF) is a promising approach to enhance sport performance, yet its effectiveness remains heterogeneous. Athletic expertise emerges as a primary candidate for this heterogeneity, potentially determining how athletes engage with different feedback designs to modulate sensorimotor rhythms (SMRs). Whether explanatory feedback (neural processes) and evaluative feedback (performance magnitude) differentially benefit athletes as a function of expertise remains unknown. We examined these designs across novice, intermediate, and expert basketball athletes imagining free throws across four feedback conditions: no feedback (None), an evaluative target (Evaluative), an explanatory topographic map (Explanatory), or both (Both). While intermediates and experts improved when feedback was present relative to None, novices showed no benefit from feedback presence alone. Crucially, gaze–performance coupling revealed a double dissociation: novice success increased with fixation on explanatory information, whereas expert success increased with fixation on evaluative information. These findings support the implementation of adaptive feedback designs to optimize MI-NF training.
• Towards an Integrated Model of Onset Detection and Multiclass RecognitionPeña, David Manuel Carmona; Gracìa, Carlos Alberto Reyes; Garcìa, Alejandro Antonio Torres; 10.3217/978-3-99161-093-9-058Brain-Computer Interfaces (BCIs) based on imagined speech offer a promising alternative for intuitive communication; however, the transition to real-time asynchronous systems is hindered by the "onset detection" problem. This study evaluates the efficacy of Instant Wavelet Energies (IWE), Empirical Mode Decomposition (EMD), and Generalized Hurst Exponent (GHE) for identifying the transition between Idle State Segments (ISS) and Imagined Word Segments (IWS). We analyzed three datasets (DS1, DS2, and DS3) using Support Vector Machines (SVM), Random Forest (RF), k-NN, and Logistic Regression (LogReg).Our results demonstrate that binary intent detection consistently outperforms multiclass word recognition, with the latter remaining a significant challenge. The MCC remains < 0.35 for most configurations, indicating that while these features are effective for intention "gating"; however, they exhibit a lack of the requisite specificity to achieve decisive discrimination.
• Using brain network features in motor imagery-based BCI for stroke rehabilitationBousfiha, Camile; Gabot, Camille; Baili, Diana; Couton, Marion; Adwane, Grace; Obadia, Michael; Zavanone, Chiara; Dupont, Sophie; Marcilliere, Quentin; Bayen, Eleonore; Bougrain, Laurent; Bartolomeo, Paolo; de Vico Fallani, Fabrizio; 10.3217/978-3-99161-093-9-059Motor imagery brain-computer interfaces are promising tools for post-stroke rehabilitation, but most systems rely on spectral power features and are often tested in highly selected patient populations. Because stroke affects large-scale brain networks, connectivitybased markers may provide complementary information for decoding and monitoring rehabilitation. This pilot study evaluated the feasibility of an intensive 6-week MIBCI protocol combined with functional electrical stimulation in five chronic stroke patients with heterogeneous and often severe motor impairment. Functional connectivity analyses revealed the emergence of motor imagery-related modulations over centro-frontal and centro-parietal regions. These findings support the feasibility of intensive MI-BCI rehabilitation in chronic stroke and suggest that functional connectivity may be a relevant biomarker for adaptive BCI strategies.
• MindRun: an Accessible Multi-Player BCI Game for Collaborative Motor Imagery TrainingPalatella, Alessio; Tortora, Stefano; Menegatti, Emanuele; Tonin, Luca; Alimardani, Maryam; 10.3217/978-3-99161-093-9-060Motor imagery (MI) brain-computer interfaces (BCIs) often suffer from limited engagement and reliance on high-cost electroencephalography (EEG) systems. We present MindRun, a continuous MI-BCI game supporting both single and multi-player interactions using a low-cost eight-channel EEG headset. The system was evaluated in a feasibility study with 10 healthy participants who engaged in both collaborative and solo game modes. EEG signals were decoded using a Riemannian geometry-based framework, yielding probabilistic outputs for continuous avatar control. In collaborative mode, the classifier combined individual outputs through a weighted probabilistic fusion to achieve stable joint control. Classification accuracy was significantly above chance level, and participants were able to achieve goaldirected steering in the game. These results demonstrate that continuous single and collaborative MI-BCI gaming is feasible with low-cost EEG hardware, highlighting the importance of interaction design and probabilistic control strategies for accessible real-time MI-BCI systems.
• Benchmarking Feature-Level and Decision-Level Fusion Strategies for Hybrid EEG-fNIRS Cognitive State DetectionCarpio-Chicote, Ainara; Deaconu, Tudor-Andrei; Gómez, Enrique J.; Kontaxakis, George; 10.3217/978-3-99161-093-9-061The integration of Electroencephalography (EEG) and functional Near-Infrared Spectroscopy (fNIRS) has emerged as a robust framework for passive Brain-Computer Interface (pBCI) applications, particularly in the field of cognitive workload monitoring. This study benchmarks feature-level (early) and decisionlevel (late) fusion strategies using a subject-independent Leave-One-Subject-Out (LOSO) validation framework. Leveraging an open-access dataset of 26 participants performing an n-back task (0-back vs. 2/3-back), we developed a MATLAB-Python pipeline to contrast unimodal and multimodal performance. Our results demonstrate that while EEG provides high temporal responsiveness, its inter-subject variability limits generalization (65.9% LOSO accuracy). In contrast, fNIRS offers stable hemodynamic markers (76.9% accuracy). Multimodal integration significantly outperformed unimodal baselines, with decision-level fusion achieving the highest subject-independent performance (85.3% accuracy, 0.977 AUC) using an SVM-Poly meta-classifier. These findings demonstrate that hybrid EEG-fNIRS integration provides a more reliable framework than unimodal approaches for detecting distinct cognitive states. While further validation is needed for granular workload levels, this study establishes a robust baseline for the development of future pBCI applications in operational environments.
• Scene-based Stimulation in the Context of a Visual ERP BCIEnkhbat, Amar; Tangermann, Michael; Thielen, Jordy; 10.3217/978-3-99161-093-9-062Visual brain–computer interfaces (BCIs) provide a promising communication channel for individuals with severe motor impairments. One well-known example is the event-related potential (ERP) matrix speller, which enables users to select characters on a computer screen. However, such systems typically require users to repeatedly shift attention between a screen-based interface and the physical environment. In this work, we propose an in-the-scene BCI paradigm, in which users interact directly with real-world objects rather than symbolic representations on a screen. We evaluated the proposed paradigm in offline experiments in which healthy participants (N = 22) selected target objects under two conditions: a screen-based BCI and an in-the-scene laser-based BCI. Comparisons of the elicited ERPs showed differences between the two stimulation conditions, including distinct scalp activation patterns and significantly larger N1 amplitudes in the in-the-scene condition. These findings indicate that neural responses elicited by real-world object stimulation differ from those obtained with conventional screen-based BCIs. This suggests that decoding methods may require adaptation for in-the-scene BCI systems.
• Connectivity Between Scalp EEG and Intracranial Recordings During 40 Hz Visual StimulationMirsaeedi, Mani; Mlinarič, Tjaša; Carrette, Evelien; Van Roost, Dirk; Boon, Paul; Meurs, Alfred; Van Hulle, Marc M.; 10.3217/978-3-99161-093-9-063We investigated whether rhythmic visual stimulation at 40 Hz reveals connectivity-based relationships between scalp EEG and intracranial recordings. Two epilepsy pa-tients with simultaneous EEG, ECoG, and SEEG record-ings participated in regular, oddball, and irregular stimu-lation conditions. Connectivity was quantified using co-herence and weighted phase lag index (wPLI). Rhythmic stimulation produced robust coupling between scalp and invasive electrodes, whereas irregular stimulation did not. These findings suggest that scalp EEG can reflect neural activity recorded in cortical and deeper brain structures during rhythmic stimulation.
• Comparison of Lightweight EEG Pipelines for Silent Speech InterfacesNikolopoulos, Emmanouil; Kontaxakis, George; Gomez, Enrique J.; Oropesa, Ignacio; 10.3217/978-3-99161-093-9-064This study investigates various machine learning (ML) and deep learning (DL) methodologies for decoding inner speech using non-invasive electroencephalography (EEG). Utilizing a publicly available high-density 128-channel dataset from ten participants, nine distinct computational scenarios were tested, ranging from traditional spatial filtering and statistical ensembles to Riemannian geometry and convolutional neural networks. The research aims to compare and identify the most effective signal processing strategies to overcome the inherently low signal-to-noise ratio of EEG signals generated during inner speech. Results demonstrate that Riemannian manifold-based approaches and compact deep learning architectures significantly outperform traditional Euclidean-based methods, achieving average accuracies across the ten participants of 32.41% and 30.48%, with peak accuracies reaching 44.50% and 34.50% respectively, compared with the 25% chance level of a four-class inner speech task.
• EEG-Based Identification of Individuals With Spinal Cord Injury Using Motor- Related and Visual Evoked PotentialsNikolopoulos, Emmanouil; Kostoglou, Kyriaki; Fidas, Christos A.; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-065A growing area of brain computer interface (BCI) research focuses on user identification, where individual neural signatures support secure authentication and personalization. Visual evoked potentials (VEPs) are widely used for their reliability, yet the integration of multiple neural modalities for identification remains underexplored. This study investigated, for the first time, the combined use of movement-related cortical potentials (MRCPs) and VEP features for electroencephalography (EEG)-based identification of individuals with spinal cord injury (SCI). We analyzed a publicly available EEG dataset from ten participants with cervical SCI who attempted five hand movements: pronation, supination, palmar grasp, lateral grasp, and hand open. VEPs were elicited by visual cues instructing the movement attempts, while MRCPs reflected the motor preparation and execution processes. Classification results showed that the “hand open” movement achieved the highest identification accuracy (40.28%, 4.03 × above the 10% chance level). These findings demonstrate that combining MRCP and cue-evoked VEP features offers a promising multimodal biometric approach for identifying individuals with SCI.
• Deep Sleep Classification via EEG Signal Criticality: A Passive BCI Approach for Sleep-Improvement NeurofeedbackNarębski, Stanisław; Komendziński, Tomasz; Rutkowski, Tomasz M.; 10.3217/978-3-99161-093-9-066Automated sleep staging is a fundamental application of passive Brain-Computer Interfaces (pBCI), decoding spontaneous neural states to enable closed-loop interventions independent of user intent. This study evaluates criticality features derived from Detrended Fluctuation Analysis (DFA) for the specific identification of deep sleep (N3). We analyzed 347,232 EEG epochs from 290 older women using UMAP manifold learning to visualize state transitions. Subsequently, six classifiers were benchmarked via 10-fold cross-validation, using balanced accuracy to determine the optimal "state-sensing" engine for neurofeedback.Naive Bayes achieved the highest mean balanced accuracy (87.17%±0.24%), significantly outperforming a fully connected deep neural network (FNN: 81.58%) and Random Forest (80.97%). Linear models (LDA: 57.21%; SVM: 51.01%) performed poorly, indicating that DFA-derived criticality features reside on a distinct, non-linear manifold. Probabilistic decoding of EEG criticality provides a highaccuracy sensing mechanism for pBCIs. This robust classification pipeline supports the development of statedependent neurofeedback, such as targeted auditory stimulation, to enhance cognitive recovery.
• Execution and Omission Errors in Human–Robot Interaction: An EEG StudyMu, Jing; Wimmer, Michael; Wimmer, Hannah S.; Burkitt, Anthony N.; Müller-Putz, Gernot R.; Grayden, David B.; 10.3217/978-3-99161-093-9-067Brain-computer interface research has extensively investigated the use of error-related potentials in human-robot interaction. However, while previous research demonstrated the successful decoding of erroneous task execution from the user’s electroencephalogram, errors of omission, i.e., the non-execution of an expected activity, remain underexplored. The current work investigates the cortical responses to both execution and omission errors during a robot-assisted task. The grand average event-related potentials (N = 12) revealed a significantly attenuated error positivity related to activity omission, likely reflecting reduced confidence in the occurrence of an error. In addition, we could decode both types of errors at a single-trial level with an average balanced accuracy of 80 % in a synchronous classification approach. The results of this work contribute to the development of autonomously corrective systems for humanrobot interaction.
• Investigating self-introspection in an auditory ERP-based BCI: A pilot experimentYang, Man; Thielen, Jordy; Tangermann, Michael; 10.3217/978-3-99161-093-9-068The decoding performance of a brain–computer interface (BCI) depends on both, the quality of the classification model and the signal-tonoise ratio of the brain signals. In this pilot study, we introduce self-introspection as a subjective measure of a user’s awareness of their own BCI performance, i.e., their ability to anticipate how well the BCI will work. We investigated this concept in an auditory event-related potential (ERP)-based BCI paradigm across six experimental sessions of a single participant. After each trial, the participant rated the expected BCI feedback using a continuous mechanical slider, before receiving the true BCI feedback. This enabled a direct comparison between subjective self-introspection scores and BCI system scores. Results indicate a positive relationship between these scores across sessions, suggesting that the user had an internal sense of his BCI performance. As the present study included a single participant, the findings should be interpreted as a feasibility demonstration. Nevertheless, they show that subjective system performance can be estimated within an auditory ERP BCI paradigm, providing a potential method for studying user learning and adaptation in BCI use.
• EEG markers of cognitive impairment in cerebral small vessel disease and neurofeedback: a narrative reviewBourdil, Manon; Sauzéon, Hélène; Lotte, Fabien; 10.3217/978-3-99161-093-9-069Cerebral small vessel disease (CSVD) is a major cause of vascular cognitive impairment (VCI) and vascular dementia (VaD), yet early detection and monitoring remain challenging. Electroencephalography (EEG) may offer a non-invasive and cost-effective approach to identify functional brain alterations associated with CSVD. This narrative review summarizes current evidence on EEG markers in CSVD patients with mild or major cognitive impairment. Across studies, CSVD was consistently associated with spectral slowing (increased delta and theta power, reduced alpha activity), disrupted functional connectivity (particularly in fronto-parietal networks) and altered event-related potentials. Several EEG measures correlated with cognitive performance and lesion burden, suggesting potential value for disease monitoring. However, heterogeneity in diagnostic criteria and methodological approaches limits comparability across studies. Overall, EEG biomarkers appear promising for early detection and monitoring of CSVD and may provide targets for future neurofeedback interventions aimed at mitigating cognitive decline, which we also discuss.
• Benchmarking Positional Encoding Strategies for Transformer-Based EEG Foundation ModelsYüce, Ayse Betül; Stober, Sebastian; 10.3217/978-3-99161-093-9-070Electroencephalography (EEG) is a widely used non-invasive technique for measuring brain activity in brain–computer interface (BCI) applications. Supervised EEG decoding models often struggle to generalize across tasks, subjects, and datasets, motivating transformer-based EEG foundation models trained with self-supervised learning. Since transformers are permutation-invariant, they require explicit positional information. Unlike textual tokens, EEG electrodes are spatially distributed across the scalp, raising the question of how electrode positions should be encoded in transformer-based EEG models. In this study, we benchmark five positional encoding strategies within the CBraMod backbone and evaluate them under linear probing and fine-tuning protocols on motor imagery classification and emotion recognition. Our results show that no single strategy consistently outperforms across tasks. Spherical Positional Encoding (SPE) yields strong representations for motor imagery but underperforms on emotion recognition, while Asymmetric Conditional Positional Encoding (ACPE) demonstrates more consistent performance across tasks. These findings suggest that the optimal positional encoding strategy is task-dependent, with no universal solution across EEG decoding scenarios.
• Gamification Improves Neurofeedback Performance and Experience among Users with ADHD: Findings from NeuroWizard IIKohnen, Henriette; Mavromoustakos-Blom, Paris; Alimardani, Maryam; 10.3217/978-3-99161-093-9-071This study explored the effect of gamification on neurofeedback (NF) training in individuals with Attention-Deficit/Hyperactivity Disorder (ADHD). In a within-subjects design, 16 participants with ADHD completed both a gamified and a non-gamified NF task delivered via a Brain-Computer Interface (BCI). Task performance and user experience were evaluated across conditions. The gamified condition led to a higher task accuracy and better user experience, characterized by greater immersion and lower negative affect. However, some participants also reported increased mental fatigue. Notably, most participants expressed a preference for the gamified version, describing it as more motivating and engaging. These findings suggest that carefully applied gamification can enhance both the effectiveness and enjoyment of NF training for individuals with ADHD, paving the way for the development of more user-centered therapeutic tools.
• Towards plug & play c-VEP-based BCIs: a pilot study on zero-calibration adaptive decoding using deep learningSantamaría-Vázquez, Eduardo; Martínez-Cagigal, Víctor; Martín-Fernández, Ana; Hornero, Roberto; 10.3217/978-3-99161-093-9-072Brain-computer interfaces (BCI) based on code-modulated visual evoked potential (c-VEP) provide fast and robust control, but most high-performance systems still require user-specific calibration before use. This work evaluates a zero-calibration decoding strategy for a 16-command c-VEP speller encoded with binary m-sequences, aiming to reduce preparation time while preserving the possibility of online subject personaliza- tion. The proposed model is based on the original EEG- Inception implementation, with task-specific adaptations. We also developed a novel adaptive strategy that simulates online operation, in which high-confidence predictions are converted into pseudo-labels to enable conservative unsupervised adaptation of the batch-normalization layers and the classification head. We evaluated the performance of our approach on 15 subjects using a leave-one-subject-out protocol. Results showed that the zero-calibration approach without adaptation achieved a mean command accuracy of 84.79% after 10 stimulation cycles (5.25s). When two outlier subjects were excluded, this value increased to 95.91%. The adaptive strategy further improved performance, particularly during the early stimulation cycles, with a mean improvement of 5.5%. These findings support the feasibility of subject-independent c- VEP decoding using deep learning and suggest that conservative test-time adaptation can improve robustness in practical plug-and-play spelling scenarios.
• Adaptive Classification of EEG Signals for Bimanual Motor Imagery using Riemannian GeometryTéllez, Ahiram Cortés; Gutiérrez, Jesús López; García, Alejandro Torres; 10.3217/978-3-99161-093-9-073Brain-computer interfaces (BCIs) experience severe performance degradation when transitioning from unilateral to bimanual decoding due to simultaneous bihemispheric interference. To address this, we developed the Adaptive Topological BCI (AT-BCI) architecture. By embedding temporal dynamics into Hankel matrices and applying Riemannian geometry to align neural distributions, the model enables zero-shot transfer without target-domain calibration. Treating the covariance shift as a purely geometric discrepancy isolates mixed signals and prevents the expansion of peripheral noise. We evaluated AT-BCI on the PhysioNet dataset, projecting unilateral motor imagery onto bimanual tasks. After applying a strict amplitude filter to exclude electromyographic artifacts and physiological desynchronization, evaluation on a validated cohort of 10 subjects yielded a mean F1-score of 0.7468 and a Cohen’s Kappa of 0.4901. These results demonstrate that restricting spatial topology and enforcing geodesic alignment stabilizes bimanual control, advancing the clinical viability of neuroprosthetic interfaces without the burden of daily recalibration.
• Co-Adaptive MRCP-based BCIs: Modulation of Cortical Potentials by Enhanced Classifier PerformanceSuwandjieff, Patrick; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-074Movement-related cortical potentials (MRCPs) reflect voluntary motor intentions and are widely used as control signals in asynchronous brain-computer interfaces (BCIs). However, the extent to which task engagement and feedback influence MRCP dynamics remains unclear. We analyzed MRCPs across two identical online speller sessions to examine how task context and classifier reliability relate to preparatory brain activity. Healthy participants controlled a row/column speller by performing right-hand gestures, first using a classifier trained on cue-aligned offline data and subsequently using a classifier retrained on data collected during the online task. Trials segmented into four task stages of increasing complexity showed no significant differences in MRCP amplitudes, suggesting limited within-session effects. In contrast, MRCP amplitudes were significantly higher in the second session following retraining. This pattern suggests an association between improved classifier performance, task success, and enhanced MRCP expression, supporting a co-adaptive interpretation and highlighting the importance of session-wise factors in MRCP-based BCI operation.
• Brain-computer interface performance development over 30 hours of patient training with an auditory event-related potential protocolTangermann, Michael; de Ranitz, Alexander; Raheem, Soz; Musso, Mariacristina; 10.3217/978-3-99161-093-9-075Neural activity measured with electroencephalography can be decoded into control commands using a brain-computer interface (BCI). Longitudinal studies on BCI training are scarce. While some studies examined long-term training on motor imagery (MI) BCIs, multi-session user training using event-related potential (ERP) protocols remains largely unexplored, despite its relevance for novel cognitive rehabilitation protocols. Methods: In a post-hoc analysis, we investigated the ERP decoding performance in ten patients with stroke-induced aphasia, who completed a 30-hour multi-session feedback-based language training. Performance was assessed using an adaptive block-Toeplitz LDA classifier and quantified by the area under the ROC curve. Results: Decoding performance varied across patients and evolved over time. Several patients improved across sessions, indicating learning, while others showed stable performance. Increases in task difficulty caused sudden performance drops, often followed by recovery and gradual improvement. Conclusion: These findings provide insights into the longitudinal dynamics of ERP-based BCI training and highlight its potential for BCI-supported rehabilitation beyond MI paradigms.
• UniSwarm-SSVEP: Dynamic Content Mapping for Efficient SSVEP-Based Swarm Robot ControlDu, Rui; Zhang, Hongxin; Wei, Mingyi; 10.3217/978-3-99161-093-9-076Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) are widely used for device control, but static one-to-one stimulus-command mapping limits scalability in swarm tasks, and simply increasing the number of stimulus targets may reduce recognition performance. This paper proposes UniSwarm-SSVEP, a swarm robot control system with a Dynamic Content Mapping Module (DCMM). By updating stimulus meanings according to the current control state, DCMM allows tens of physical targets to be mapped to hundreds of logical commands and supports flexible switching among single-vehicle, subgroup, and whole-swarm control modes. To validate this mechanism, we constructed a 20-target UniSwarm-SSVEP test system for a 153-command swarm control task. Under FBCCA, UniSwarm-SSVEP achieved an average accuracy of 91.56% and a standard information transfer rate (ITR) of 143.46 bits/min. Task-discriminant component analysis (TDCA)-based validation further improved the accuracy to 95.78% and the standard ITR to 268.80 bits/min, demonstrating the scalability and decoding compatibility of the proposed system.
• Influence of Median Nerve Stimulation Intensity on ERD/ERS, BCI Performance and User ExperienceBechon, Loïc; Lotte, Fabien; Fleck, Stéphanie; Rimbert, Sebastien; 10.3217/978-3-99161-093-9-077Median Nerve Stimulation (MNS) has recently emerged as a promising approach to enhance Motor Imagery (MI)-based Brain–Computer Interfaces (BCIs) by providing additional somatosensory input that modulates sensori-motor rhythms. While previous studies have shown that MNS interacts with MI-related activity and can improve BCI performance, the influence of stimulation intensity on the resulting cortical patterns remains largely unexplored. In this study, we investigate how different MNS intensities affect ERD/ERS patterns, offline classification performance, and user experience (UX) with these stimulations. Thirteen participants performed right-hand motor imagery while receiving three stimulation levels relative to their motor threshold: low (below threshold), medium (twitch-inducing), and high (above threshold). Results show similar ERD/ERS patterns across stimulation intensities and no significant differences in classification accuracy. Participants reported very low levels of discomfort associated with the MNS. These findings suggest that relatively low-intensity MNS is sufficient to elicit detectable sensorimotor responses while improving user comfort, supporting the design of more usable MNS-based BCI systems. This opens the possibility of using MNS at the lowest effective level, which is particularly beneficial from a user experience perspective for prolonged BCI interaction.
• Mission Impossible? Inhibiting Frontal Midline Theta with NeurofeedbackPfeiffer, Maria; Botrel, Loïc; Kübler, Andrea; 10.3217/978-3-99161-093-9-078Frontal midline theta (FMT) is an oscillation closely linked to cognitive control and performance monitoring. A recent meta-analysis suggests that FMT can be upregulated via neurofeedback (NF), yet successful downregulation remains largely unsubstantiated and has primarily been tested in single-session designs. We tested whether multi-session training enables FMT inhibition via NF. Five participants completed six NF sessions targeting FMT reduction (4-8 Hz at Fz). Contrary to the training goal, FMT increased during NF relative to the pre-session resting-state baseline. A similar pattern was observed in a previous pilot study using individualized FMT peak-frequency NF. Together, these findings suggest that traditional NF approaches may face fundamental difficulties in achieving reliable FMT downregulation.
• Pre-Experiment Resting Baseline Improves Motor Imagery BCI Decoding in a Robotic Hand Rehabilitation ContextCruz, Aniana; Pires, Gabriel; 10.3217/978-3-99161-093-9-079Motor imagery (MI)-based brain-computer interfaces (BCIs) rely on event-related desynchronization and synchronization (ERD/ERS), which are defined relative to a resting-state baseline. Therefore, baseline selection directly affects both the interpretation of the ERD/ERS patterns and MI classification performance. In this work, we systematically evaluate the impact of baseline definition and temporal window selection on ERD/ERS patterns and classification performance in a post-stroke rehabilitation protocol involving a robotic hand exoskeleton, considering both calibration and online sessions. Three resting-state baselines were compared: a long rest recorded before the experiment, a long rest recorded after task execution, and a short inter-trial rest commonly used in MI-BCI paradigms. The results show that the pre-experiment resting baseline yielded the strongest and most stable ERD patterns, as well as the highest classification performance in both calibration and online data. Rest recorded after task execution resulted in attenuated ERD and slight performance degradation. The short inter-trial rest led to substantial variability and performance loss.
• Decoding Meaning: Cluster-Based Analysis of Semantic Processing Patterns in EEGPodmore, Joshua; Radman, Moein; Paulmann, Silke; Poli, Riccardo; Daly, Ian; 10.3217/978-3-99161-093-9-080Conventional brain-computer interfaces (BCIs) typically rely on slow character-by-character selection mechanisms. Semantic BCIs (sBCIs) aim to overcome this limitation by decoding entire concepts directly from neural activity. Here, we introduce GRASP (Game of Retrieval and Semantic Processing), a novel auditory paradigm designed to isolate volitional semantic processing in EEG recordings. Participants (n = 9) identified concrete concepts during a gamified word-guessing task and provided response-locked markers (spacebar press). Importantly, 25% of trials consisted of phonologically matched pseudo-word sentences, serving as an active baseline condition. Time-frequency representations were computed using Morlet wavelets and statistically evaluated using threshold-free cluster enhancement permutation testing to characterise the spatio-spectro-temporal signatures of semantic processing within individual participants. As an initial proof-of-concept analysis, we focus on the body vs. dummy contrast within the GRASP paradigm. Our findings provide insight into the timing, scalp regions, and frequency bands associated with EEG-based semantic processing and may support the development of targeted preprocessing and machine-learning approaches for sBCIs.
• Investigating Human and Classifier-Level Uncertainties: An EEG Study Using Grasp Imitation TaskOguz, Osman Cagri; Kasaei, Hamidreza; Frey, Steffen; Sburlea, Andreea Ioana; 10.3217/978-3-99161-093-9-081In BCI research, uncertainty quantification recently attracted researchers to enhance human-machine collaboration and transparency of the systems. Grasping constitutes one of the most crucial interactions with the environment, for which the associated risks of overconfident but inaccurate classification are high. However, there is yet to be an explicit paradigm design that allows investigating if classifier uncertainties align with that of humans. By developing a grasp imitation paradigm with visually degraded videos during the observation stage, we aimed to incur uncertainties on the grasping decisions and execution. EEG and EMG data from 8 participants is collected. The results demonstrated that among the grasp types, 5-finger precision grasp is the best classified one, with 47% within-class accuracy according to an LDA classifier using movement related cortical potentials as features. Interestingly, for the participant having highest ROC-AUC score, subjectively reported and the LDA classifer uncertainty aligns. As the data collection is still ongoing, future directions are also discussed.
• A Comparative Study of EEG Representations and Channel Settings for EEG-to-Text Decoding during Reading TasksPérez-Hernández, Eduardo V.; Torres-García, Alejandro A.; Villaseñor-Pineda, Luis; 10.3217/978-3-99161-093-9-082Decoding natural language from electroencephalography (EEG) signals remains a major challenge in brain–computer interface research for assistive communication. This work presents a comparative study of EEG representations and channel configurations for EEG-to-text decoding during reading tasks. We evaluate Hilbert–Huang Transform (HHT) features, power spectral density (PSD), and temporal statistical descriptors across three spatial configurations: full-head EEG, a reduced 10–20 system, and a Broca–Wernicke sub-set. Experiments are conducted on the ZuCo v1 and v2 datasets using fully autoregressive decoding without teacher forcing. Results show that decoding performance varies across EEG representations and channel configurations. Temporal statistical features achieve the best overlap-based performance under the full-head configuration (BLEU-1 = 11.45, ROUGE-1 F1 = 11.52), while reduced channel subsets remain competitive despite lower spatial dimensionality. These findings highlight the importance of representation design, spatial selection, and task consistency for EEG-to-text decoding under short and variable EEG segments.
• Evaluating the impact of electrode type and montage on classification performance in Assistive P300-based BCIsCaracci, Valentina; Riccio, Angela; Petrinotti, Serena; Bigioni, Alessandra; Boesso, Simone; Toppi, Jlenia; Cincotti, Febo; Mattia, Donatella; 10.3217/978-3-99161-093-9-083Severe motor disabilities from neurological conditions often prevent individuals from accessing modern Assistive Technologies (AT) for communication. Brain-Computer Interfaces (BCIs) bypass motor pathways, translating brain activity into device control. Among non-invasive BCIs, P300-based systems, exploiting the P300 Event-Related Potential, are widely investigated for communication: in the P300-speller users select letters by focusing on target stimuli, enabling communication for individuals with severe motor impairments. BCIs are therefore promising, but usability, setup complexity, and gel-based EEG caps limit integration. Traditional gel-based caps provide high-quality signals but are uncomfortable and slow to set up, while semi-dry headsets improve comfort and usability, but typically have suboptimal montages for P300-applications. This raises questions about the effects of electrode type and montage. This study compared a commercial semi-dry 7-channel headset for out-of-the-lab applications and a traditional wet EEG cap during P300-speller tasks. Classification performance was evaluated online and offline, providing insight into how electrode design and configuration influence the usability and effectiveness of BCIs.
• First Real-Time Use of Quasi-Movements for Exoskeleton Control: Comparison with MI-BCI in Healthy Participants and Pilot Application in Stroke RehabilitationYashin, Artem S.; Ivanov, Leonid A.; Kondur, Anna A.; Bobrov, Dmitry A.; Slyunkova, Elena V.; Gabuzov, Grigory G.; Lukyanov, Evgeny S.; Shevtsova, Yulia G.; Vasilyev, Anatoly N.; Bobrov, Pavel D.; Kotov, Sergey V.; Shishkin, Sergei L.; 10.3217/978-3-99161-093-9-084Quasi-movements (QM) are movement attempts minimized to produce no overt movement or muscle activation. Unlike kinesthetic motor imagery (MI), they may not require mastering an abstract mental skill; unlike conventional attempted movements, they may reduce risks of abnormal motor synergies. QM- based BCI training may therefore be a useful clinical alternative for sensorimotor BCI rehabilitation. We evaluated QM for real-time control of a hand- exoskeleton BCI. First, 19 healthy participants completed QM- and MI-based control sessions with online three-class EEG classification and post-session ratings. QM achieved performance comparable to MI in online accuracy and subjective ratings, although it required closer muscle-tension monitoring. Preferences for QM versus MI were split evenly, indicating inter- individual variability. Second, four chronic stroke patients completed repeated QM-BCI training over multiple days. Patients quickly understood the instructions and produced classification accuracies consistently above a nominal three-class chance reference. Clinical changes were heterogeneous but generally favorable. QM may therefore provide a promising additional strategy for exoskeleton-based BCI rehabilitation.
• Volitional and Non-Volitional Motor Tasks Encode Distinct EEG Dynamics: Implications for Paradigm-Specific BCI Decoder DesignMorales-Magallón, Fernando; Bojorges-Valdez, Erik; Gonzáles-Gonzáles, Daniela; 10.3217/978-3-99161-093-9-085This work explores differences in brain dynamics during the execution of voluntary and non-voluntary knee-joint movements (using an isokinetic dynamometer) by analyzing distance matrices constructed from a d-dimensional reconstructed phase-space of in- stantaneous band-power time series (iBPts). Two condition-specific decoders were developed: a volitional (Vol) model, calibrated exclusively on active motor execution, and a non-volitional (NVol) model trained on machine-assisted movement. The nonlinear features derived from the phase-space distance matrix successfully encode condition-specific dynamical structures, yielding classification metrics consistent with current BCI standards, even for data acquired across a several-day interval. The Vol model achieved higher performance when classifying voluntary data (AUC: 0.772 ±0.128) than non-voluntary data (AUC: 0.613 ±0.159), whereas the NVol model demonstrated consistent performance across both conditions (AUC: 0.694 ±0.119 for voluntary and AUC: 0.697 ±0.100 for non-voluntary). A Linear Discriminant Analysis (LDA) classifier, combining the outputs of both models, revealed condition-dependent decoder contributions, confirming that volitional and non-volitional motor states elicit separable dynamical regimes in the EEG signal. These results indicate that although transfer learning from overt motor execution enables reliable classification of non-voluntary data, the underly- ing dynamical differences lead to degraded performance. Consequently, while calibration on overt movement may result in reduced accuracy in covert scenarios, the use of covert-movement data may provide more representative performance metrics for online implementations.
• Goal-Directed Quasi-Movements Based on Natural Hand Actions: EEG Correlates and Relevance for Rehabilitation BCISvirin, Evgeniy P.; Berdyshev, Daniil A.; Shishkin, Sergei L.; 10.3217/978-3-99161-093-9-086Quasi-movements (QMs) are movements minimized to the point that they produce no visible movement or detectable muscle activation. Unlike kinesthetic motor imagery (MI), they are closer to actual movement in the sense of intention and may therefore provide a more natural basis for sensorimotor BCI control. However, previous QM research has been largely limited to thumb abduction. Here we investigated, for the first time, QMs based on more natural goal-directed hand actions, pressing and pointing. Healthy participants performed kinesthetic MI, non-goal-directed QMs, goal-directed QMs, and real movements while EEG and a custom force-sensor signal were recorded. Both QM conditions elicited stronger contralateral μ- rhythm desynchronization than MI, whereas ipsilateral μ- ERD did not differ reliably between conditions. Goal-directed instruction did not produce an additional detectable effect, and residual movement did not account for the EEG differences. These preliminary findings suggest that natural-action QMs are feasible and may represent a promising alternative to MI for non-invasive BCI applications.
• HD-EMG-Based BCI for Control and Assessment in Post-Stroke Hand Paresis: A Proof of ConceptVelardi, Gian Maria; Le, Thien N.; Cimolato, Andrea; Fasching, Bernhard; Serles, Wolfgang; Raspopovic, Stanisa; 10.3217/978-3-99161-093-9-087Brain–computer interfaces (BCIs) offer a promising avenue for restoring motor function after stroke, but conventional BCIs are often limited by lengthy calibration or invasive implantation. High-density electromyography (HD-EMG) enables non-invasive decomposition of surface muscle signals into individual motor unit (MU) spike trains, providing a direct readout of neural drive to muscle, even when corticospinal transmission is impaired (e.g., after stroke). However, the blind source separation algorithms used for decomposition typically require approximately 15 seconds of sustained contraction, a demand that many stroke survivors cannot meet due to post-stroke fatigue and reduced motor endurance. Here we present a new HD-EMG processing pipeline that extracts movement intention through motor unit decomposition using repeated brief contractions rather than a single sustained effort. We evaluate this calibration strategy in both simulated signals (with ground-truth MU activity) and stroke-participant recordings. Our repeated-contraction approach maintains the same number validated motor units compared to sustained contraction (~6.75–6.95 MUs), indicating that our approach preserves decomposition sensitivity of state of art approaches. Furthermore, using two selectively recruited motor units, we demonstrate real-time classification capability of two distinct hand movements with high accuracy. Importantly, we identify a satellite potential in the motor unit action potential waveforms. This is a biomarker of corticospinal reorganization not previously observed with non-invasive recordings of stroke patients. These findings establish a clinically feasible calibration paradigm for HD-EMG-based BCIs in neurological populations and suggest that motor unit analysis may simultaneously serve both real-time control and longitudinal monitoring of recovery.
• Single case study of Spatiotemporal Dynamics of Cortical Preparation and Motor Execution Across Lower-Limb Movements of Varying ComplexityPathak, Shreyansh; von der Linde, Lilian; Gloria, Joeé Jesus Hernandez; Ofner, Patrick; Mrachacz-Kersting, Natalie; 10.3217/978-3-99161-093-9-088Most lower-limb brain-computer interface (BCI) studies focus on either isolated single-joint movements or complex movements like gait initiation and walking, leaving an unexplored gap between the transition from one movement type to another. As part of the REWIRING project, we aim to investigate the spatiotemporal relationship between cortical signals and muscle activation across various motor tasks of increasing biomechanical complexity and different volition types. To accomplish this, in an ongoing pilot study, electroencephalographic and electromyographic data were recorded and analysed from a single participant during three self-paced movements of varying complexity: dorsiflexion, plantar flexion, and gait initiation. A preliminary analysis of the spatially-filtered data extracted from Cz indicates identifiable movement related cortical potential (MRCP) patterns among all investigated tasks, with differences in peak-negativity and waveform morphology. Our findings demonstrate the feasibility of the proposed setup for investigating and understanding the differences of MRCPs across various lower-limb movements in future studies.
• Combining Mindfulness and Implicit Neurofeedback to Upregulate Frontal Alpha AsymmetryPercolla, Elisa; Drapart, Saskia; Kübler, Andrea; Pfeiffer, Maria; 10.3217/978-3-99161-093-9-089This contribution presents an experimental design that integrates an implicit auditory neurofeedback protocol in a mindfulness intervention to pursue an upregulation of frontal alpha asymmetry (FAA). The following sections detail the design of such an experiment, together with the development and implementation of the materials in use. Lastly, preliminary data are provided to support its feasibility and discuss future improvements and applications.
• Beyond Accuracy: User Experience Matters in the Comparison of Two EEG Headsets for BCIPetrinotti, Serena; Bigioni, Alessandra; Caracci, Valentina; Boesso, Simone; Toppi, Jlenia; Cincotti, Febo; Mattia, Donatella; Riccio, Angela; 10.3217/978-3-99161-093-9-090Brain-Computer Interfaces (BCIs) are a promising Assistive Technology (AT) for individuals with severe motor impairments, enabling communication and interaction through neural activity without requiring muscular control. However, their integration into everyday assistive services remains limited, partly due to usability issues related to EEG acquisition devices. In particular, the comfort and acceptability of EEG headsets are key factors for long- term adoption. This study compared users’ subjective experience with two EEG headsets during a P300-based BCI spelling task: a gel-based cap (EasyCap), commonly used in laboratory and clinical settings, and a mobile semi-dry system (X.on) designed for out-of-lab applications. Ten healthy volunteers completed two sessions, performing the same task with each headset. After each session, participants filled out a questionnaire assessing satisfaction with headset, along with Visual Analogue Scales (VAS) evaluating perceived physical discomfort. Results showed higher overall satisfaction with the semi-dry headset, suggesting greater suitability for everyday BCI communication.
• First real-time study of a BCI based on quasi-movements: Comparison to the MI BCI within a new hybrid eye–brain–computer interface paradigmShevtsova, Yulia G.; Yashin, Artem S.; Svirin, Evgeniy P.; Vasilyev, Anatoly N.; Shishkin, Sergei L.; 10.3217/978-3-99161-093-9-091We report the first real-time study of a brain–computer interface (BCI) based on quasi- movements (QM) and compare it with standard motor imagery (MI) within a novel hybrid eye–brain– computer interface (EBCI). Eighteen healthy participants used gaze to select on-screen objects and then activated EEG-based control to interact with a game requiring precise initiation and termination of sensorimotor-task execution. Online classification performance was comparable across conditions: mean balanced accuracy was 74.25% for QM and 74.63% for MI, with 13 of 18 participants exceeding 70% in each condition. No significant differences were found between QM and MI in classifier performance, game interaction measures, or subjective ratings. Nevertheless, marked interindividual variability was observed, with some participants clearly favouring one condition over the other. These findings demonstrate that QM can support stable online BCI control in a demanding hybrid gaze–EEG paradigm and may serve as a practical alternative to MI, potentially broadening accessibility of active sensorimotor BCIs.
• Contrast-dependent decoding performance in classic and grating c-VEP stimuliEr, Sena; Schipper, Donja; Qin, Yuzhen; Thielen, Jordy; Tangermann, Michael; 10.3217/978-3-99161-093-9-092Designing effective visual brain–computer interfaces (BCIs) requires careful selection of stimulus parameters that shape the strength and reliability of evoked neural responses. Stimulus contrast is one such parameter in code-modulated visual evoked potential (c-VEP) BCIs, yet its effects on decoding performance and reliability remain incompletely characterized. Here we investigate how stimulus contrast influences decoding performance in a c-VEP speller using two stimulus families: classic luminance flicker and grating-based textured stimulation. Four participants were tested across six contrast lev- els (10–100 %). Trial-level performance was quantified using a Fisher-transformed margin metric and modeled with Gaussian processes. In this exploratory dataset, performance generally increased with contrast, with larger decoding margins for the classic stimulus than for the grating stimulus. The magnitude and shape of these effects varied across participants, while estimated trial-to-trial noise changed only modestly across contrast lev- els. These findings suggest that trial-level Gaussian-process modeling can support exploratory characterization of stimulus-parameter effects in c-VEP BCIs.
• Decoding Neuroethics: Quality as the Code for Neuro-AI GovernanceKalugina, Ekaterina; 10.3217/978-3-99161-093-9-093AI-powered brain-computer interfaces (BCIs) are rapidly advancing, offering transformative potential in healthcare and human augmentation. However, AI's ability to decode brain signals raises profound concerns about neuroethical risks and potential harms. When such systems produce erroneous outputs, it is not merely a software defect but a potential violation of cognitive liberty, mental privacy, and neural integrity. The dominant governance response – neuroethics – has largely focused on rights articulation rather than quality as the operational mechanism capable of translating ethical principles into enforceable controls. This gap is compounded by a structural accountability problem across the BCI value chain. While AI capabilities evolve dynamically, existing safety frameworks remain static, and often do not keep pace with technological developments. This article presents a normative doctrinal analysis of EU product safety regulations as they apply to AI-powered BCIs. It examines how quality obligations, risk assessment, and liability should be structured across the BCI value chain in response to the neuroethical risks posed by neural decoding systems. On that basis, the article argues that quality should function as the central legal standard for AI-powered BCIs and proposes a dynamic NeuroAI Governance Framework to operationalize that position.
• Epidural micro-ECoG provides a robust interface to the cortex over 4 years in non-human primatesGkogkidis, Constantin Alexis; Schuettler, Martin; Ulloa, Miguel; Ryou, Jae Wook; Purpura, Keith P.; Baker, Jonathan Lee; 10.3217/978-3-99161-093-9-094Implantable brain-computer interfaces (BCIs) are on the verge of transitioning from the research and investigational phase to clinical readiness stage. Micro-electrocorticography (µECoG) is one possible choice of recording modality and shows promising longterm results in both epi- and subdural implantations with direct clinical translational impact. In this paper, we present longitudinal data from epidural implantation in non-human primates. The results show that neural patterns stay comparable over the course of multiple years, and that task- and condition-specific cortical signal dynamics can be distinguished. The findings satisfy two of the most important criteria for µECoG-based chronic BCIs: long-term stability of neural signal quality and the presence of potentially decodable information.
• Hacking BCI EthicsWelter, Marc; Lotte, Fabien; 10.3217/978-3-99161-093-9-095In the age of surveillance capitalism, the exploitation of personal data has become ubiquitous. Although, this can be beneficial to personalize user experience, such data is also regularly exploited for manipulation and harm, e.g. to influence voting behaviour or even incite violence. As the emergence of Brain-Computer- Interface (BCI) technologies can grant unprecedented access to private and mental data, BCI users need to be pro- tected against unethical exploits of their brain data. To be able to defend against attacks, it is helpful to be aware of possible attacks a priori. To this end, we propose to adopt an adversarial approach inspired by ethical hacking in cybersecurity to hack BCI ethics. By taking the hypothetical perspective of unethical actors, e.g. surveillance capitalists, we can determine concrete ethical vulnerabilities in BCI systems and work on patches to fix them.
• Low-Power and Compact Time-Domain Analog Feature Extraction for ECoG Motor Intent DecodingJaoui, Ilan; Saad, Joe; Hardy, Emmanuel; Struber, Lucas; Aksenova, Tetiana; Badets, Franck; 10.3217/978-3-99161-093-9-096Power consumption has become a key constraint in designing implantable Brain-Computer Interfaces (BCIs). For systems with a large number of channels, analog-to-digital conversion and Feature Extraction (FEx) are often the most power-hungry stage of the decoding pipeline. To address this bottleneck, we present a BCI processing chain using an ultra low-power compact analog Time-Domain (TD) Band-Pass Filter (BPF) bank to extract features from ElectroCorticoGraphy (ECoG) signals. First, we show that the frequency spread of TD BPF have little impact on the decoding accuracy. The proposed analog features are then assessed on a 5-class motor decoding task and achieve a decoding performance well above chance level, indicating that these features preserve discriminating information for multi-class decoding. The same features are then used for binary motor/idle detection task and performed as well as digital features, while consuming ×646 less power, and ×75 vs analog counterparts. In this setting, we also show that our system maintains robust accuracy even at high noise levels, highlighting a favorable trade-off when using lower power Analog ReadOuts (AROs). This paper overall demonstrates that TD Analog Feature Extraction (AFEx) can provide a good power-performance trade-off for binary tasks in low-power implantable BCI systems.
• Beyond Hemispheric Laterality: Riemannian Metrics as Markers of Motor Imagery Performance in BCI TrainingMurali, Sainath; Sudeep, Aryan; Lakshminarayanan, Raghavendran; Thirugnanasambandam, Nivethida; 10.3217/978-3-99161-093-9-097A major challenge in motor imagery based brain-computer interfaces (MI-BCIs) is BCI illiteracy, where many users fail to achieve reliable control despite training. Most MI-BCI decoding approaches rely heavily on hemispheric laterality of sensorimotor rhythms, overlooking a key aspect of BCI learning - the user’s ability to generate distinct and stable neural representations during MI. Here, we aimed to identify a quantifiable neural marker of MI by examining whether neural dynamics captured by Riemannian geometry–based EEG metrics relate to users’ self-reported MI performance across runs. Participants were categorized as Improved, Worsened, or Unchanged based on perceived BCI accuracy across multiple runs, and three covariance-based metrics - class stability, class distinctiveness, and rest distinctiveness were estimated. Our results show that these metrics are modulated differently across participant groups, suggesting that MI decoding reflects multidimensional neural dynamics beyond hemispheric laterality and that Riemannian metrics may help assess an individual’s ability to generate effective MI signals for BCI control.
• Physiological-Based Change Point Detection During Sustained Attention: A multimodal EEG and Eye-Tracking StudyThomas, Naomi; Al-Mfarej, Dalya; Tung, James; 10.3217/978-3-99161-093-9-098Distinguishing adaptive cognitive effort from fatigue in real-time is critical for optimising task difficulty in motor rehabilitation. This study applies a data-driven change-point detection framework to 60- minute Attention Network Task (ANT) recordings from 7 healthy participants to detect cognitive state transition by integrating spectral and connectivity EEG features, and eye-tracking features. For each participant, 7 classifiers were trained at every candidate split time to identify the temporal boundary at which pre- and post-transition physiological and behavioural states were maximally discriminable. Multimodal fusion of physiological features achieved a mean ROC AUC of 0.933 ± 0.050, exceeding behavioural features alone by 0.245 AUC units (t (6) = 17.89, p < 0.001) on average. The grand mean physiological-based split occurred at 21.6 ± 7.4 minutes, with substantial inter-individual variability (range: 13.4– 35.0 min). Post-split exhibited decreased α /αratios, increased α power, and pupil enlargement, consistent with compensatory effortful processing rather than passive fatigue. The differ-ence between EEG and behavioural based change-point time (∆ = 2.9 ± 14.4 minutes) demonstrates that behavioural monitoring alone cannot detect when neural state transitions occur, underscoring the need for individ-ualised physiological monitoring within Challenge Point Framework-based rehabilitation systems.
• Evaluating the Impact of Generic Model Priors on User-Specific Calibration in P300 Brain–Computer InterfacesGupta, Arnav; Mainsah, Boyla O.; Collins, Leslie M.; 10.3217/978-3-99161-093-9-099The current standard approach for training brain-computer interface (BCI) models is user-specific. Due to high inter-individual variability, there is often a performance gap between generic BCI models trained on data from other users relative to user-specific models. Alternatively, generic models can provide informative priors to reduce BCI calibration time. We investigate the impact of model priors on shrinkage linear discriminant analysis (LDA) with Bayesian sequential updates using a normal-inverse-Wishart (NIW) conjugate prior. Baseline WD and CD P300 classifiers achieved off-the-shelf performance of 0.629 ±0.022 and 0.612 ±0.020 area under the curve (AUC), respectively. Incorporating 20% user-specific calibration data (total = 3780 samples) with WD and CD priors yielded a statistically significant increase to 0.683 ±0.022 AUC with convergence of both models to user-specific performance 0.699 ±0.023. Sequential updates of shrinkage LDA (20% data increments) required on average 0.001 seconds with WD/CD priors vs. 0.140 seconds with only user-specific data, demonstrating that generic informative priors efficiently handle domain shifts with reduced computational overhead.
• Similar morphology, different timing: Cortico-spinal response delays following median and ulnar nerve stimulationOberndorfer, Markus E.; Golser, Bernhard; Müller-Putz, Gernot R.; 10.3217/978-3-99161-093-9-100Understanding the temporal propagation of sensory signals from peripheral nerves to the central nervous system remains challenging due to limited access to spinal cord activity. In this study, we investigate the capability of electrospinography (ESG) to resolve conduction delays following transcutaneous stimulation of the ulnar (UNS) and median (MNS) nerves. ESG and electroencephalography (EEG) recordings were obtained, preprocessed, and analyzed using latency estimation and cross-correlation methods. Spinal responses exhibited condition-specific latency differences, with UNS stimulation consistently preceding MNS stimulation by approximately 2–3 ms, in agreement with delays predicted from peripheral nerve anatomy and conduction velocity. Comparable delay patterns were observed in cortical EEG responses after averaging across participants, and waveform morphology remained similar between conditions. These findings suggest that ESG provides temporally precise information on ascending sensory signal propagation that may not be detectable at the cortical level alone. The results highlight the potential of ESG as a complementary tool for studying spatiotemporal dynamics within the human sensorimotor system.
• Volitional neural control with the CorTec Brain Interchange by an individual with chronic strokeSarma, Devapratim; Cho, Hanbin; Robert-Gonzalez, Adrià; Stephan, Emma; Schuettler, Martin; Buchheit, Marina; Gkogkidis, Constantin Alexis; Herron, Jeffrey; Weaver, Kurt E.; Cramer, Steven C.; Moritz, Chet; Ojemann, Jeffrey G.; Bahadori-Nejad, Maryam; 10.3217/978-3-99161-093-9-101We report the first demonstration of volitional control of a Brain-Computer Interface (BCI) using the CorTec Brain Interchange by an individual with chronic hemiparetic stroke and no prior BCI experience. Subdural electrocorticographic (ECoG) recordings were acquired from sensorimotor cortex to facilitate voluntary control over a velocity-defined Pong paddle. Lineardiscriminant and autoregressive ridge-regression decoders trained on mu and beta band-power features were trained on data collected during monthly study visits prior to the BCI testing. Across three short Pong sessions (96 min; 400 rallies) the participant attained 0.53–0.61 mean paddle hits per rally, above the static- paddle physics baseline (0.28; permutation p ≤ 0.013 for all three sessions) and a stricter pipeline-replay shuffled baseline (~0.36; p ≤ 0.011). Participant self-report demonstrated strategy-level insight into their control of motor-imagery and related game performance.