The Spatio-Temporal Equalization Sliding-Window Distribution Distance Maximization Based on Unsupervised Learning for Online Event-Related Potential-Based Brain–Computer Interfaces

Machines, Год журнала: 2025, Номер 13(4), С. 282 - 282

Опубликована: Март 29, 2025

Brain–computer interfaces (BCIs) provide a direct communication pathway between the central nervous system and external environments, enabling human–machine interaction control. Among them, event-related potential (ERP)-based BCIs are among most accurate reliable BCI systems. However, current mainstream classification algorithms struggle to eliminate calibration requirements rely heavily on costly labeled data, limiting practical usability of ERP-based BCIs. To address this, development unsupervised is critical for advancing real-world applications. In this study, we propose spatio-temporal equalization sliding-window distribution distance maximization (STE-sDDM) algorithm, which introduces (STE) ERP first time integrates it with novel method, (sDDM). STE estimates removes colored noise interference in background enhance signal-to-noise ratio inputs sDDM. Meanwhile, sDDM leverages an enhanced inter-class divergence metric based ergodic hypothesis theory, utilizing sliding windows emphasize temporally discriminative features, thereby improving accuracy. The experimental results demonstrate that integration significantly enhances feature separability, outperforming state-of-the-art online spelling accuracy information transfer rate (ITR), facilitating more faster plug-and-play real-time control Additionally, static equalizer architectures were found outperform dynamic when combined framework.

Язык: Английский

---

Dual-modal flexible sensors based on flexible Ti3C2Tx (MXene)-bacterial cellulose composites for neural network-assisted pronunciations, shapes, and materials perception

Journal of Alloys and Compounds, Год журнала: 2025, Номер unknown, С. 180095 - 180095

Опубликована: Март 1, 2025

Язык: Английский

---

Advances in brain computer interface for amyotrophic lateral sclerosis communication

Brain‐X, Год журнала: 2025, Номер 3(1)

Опубликована: Март 1, 2025

Abstract Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease that often results in the loss of speech, creating significant communication barriers. Brain–computer interfaces (BCIs) provide transformative solution for restoring and enhancing quality life ALS individuals. Recent advances implantable electrocorticographic systems have demonstrated feasibility synthesizing intelligible speech directly from neural activity. By recording high‐resolution signals motor, premotor, somatosensory cortices with decoding algorithms, these can transform patterns into acoustic features providing natural intuitive pathways Non‐invasive electroencephalography, while lacking spatial resolution systems, offers safer alternative high temporal capturing speech‐related dynamics. When combined robust feature extraction techniques, such as common pattern time‐frequency analyses, well multimodal integration functional near‐infrared spectroscopy or electromyography, it effectively enhances accuracy system robustness. Despite progress, challenges remain, including user variability, BCI illiteracy, impact fatigue on performance. Personalized models, adaptive secure frameworks brain data privacy are essential addressing limitations, enabling BCIs to enhance accessibility reliability. Advancing technologies methodologies holds immense promise independence bridging gap individuals ALS. Future research could focus long‐term clinical studies evaluate stability effectiveness development more unobtrusive paradigms.

Язык: Английский

---

Toward brain-computer interface speller with movement-related cortical potentials as control signals

Frontiers in Human Neuroscience, Год журнала: 2025, Номер 19

Опубликована: Апрель 2, 2025

Brain Computer Interface spellers offer a promising alternative for individuals with Amyotrophic Lateral Sclerosis (ALS) by facilitating communication without relying on muscle activity. This study assessed the feasibility of using movement related cortical potentials (MRCPs) as control signal Brain-Computer speller in an offline setting. Unlike motor imagery-based BCIs, this focused executed movements. Fifteen healthy subjects performed three spelling tasks that involved choosing specific letters displayed computer screen performing ballistic dorsiflexion dominant foot. Electroencephalographic signals were recorded from 10 sites centered around Cz. Three conditions tested to evaluate MRCP performance under varying task demands: condition repeated selections letter "O" isolate movement-related brain activity; phrase structured text ("HELLO IM FINE") simulate meaningful moderate cognitive load; and random randomized sequence introduce higher complexity removing linguistic or semantic context. The success rate, defined presence MRCP, was manually determined. It approximately 69% both conditions, slight decrease condition, likely due increased complexity. Significant differences features observed between Laplacian filtering, whereas no significant found single-site Cz recordings. These results contribute development MRCP-based BCI demonstrating their task. However, further research is required implement validate real-time applications.

Язык: Английский

---

Research on Adaptive Discriminating Method of Brain–Computer Interface for Motor Imagination

Brain Sciences, Год журнала: 2025, Номер 15(4), С. 412 - 412

Опубликована: Апрель 18, 2025

(1) Background: Brain–computer interface (BCI) technology represents a cutting-edge field that integrates brain intelligence with machine intelligence. Unlike BCIs rely on external stimuli, motor imagery-based (MI-BCIs) generate usable signals based an individual’s imagination of specific actions. Due to the highly individualized nature these signals, identifying individuals who are better suited for MI-BCI applications and improving its efficiency is critical. (2) Methods: This study collected four imagery tasks (left hand, right foot, tongue) from 50 healthy subjects evaluated adaptability through classification accuracy. Functional networks were constructed using weighted phase lag index (WPLI), relevant graph theory parameters calculated explore relationship between functional networks. (3) Results: Research has demonstrated strong correlation network characteristics tongue adaptability. Specifically, nodal degree characteristic path length in hemisphere found be significantly correlated accuracy (p < 0.05). (4) Conclusions: The findings this offer new insights into mechanisms imagery, suggesting holds potential as predictor

Язык: Английский

---

Probing Nanotopography-Mediated Macrophage Polarization via Integrated Machine Learning and Combinatorial Biophysical Cue Mapping

ACS Nano, Год журнала: 2024, Номер 18(37), С. 25465 - 25477

Опубликована: Сен. 3, 2024

Inflammatory responses, leading to fibrosis and potential host rejection, significantly hinder the long-term success widespread adoption of biomedical implants. The ability control investigated macrophage inflammatory responses at implant-macrophage interface would be critical for reducing chronic inflammation improving tissue integration. Nonetheless, systematic investigation how surface topography affects polarization is typically complicated by restricted complexity accessible nanostructures, difficulties in achieving exact control, biased preselection experimental parameters. In response these problems, we developed a large-scale, high-content combinatorial biophysical cue (CBC) array enabling high-throughput screening (HTS) effects nanotopography on subsequent processes. Our CBC array, created utilizing dynamic laser interference lithography (DLIL) technology, contains over 1 million nanotopographies, ranging from nanolines nanogrids intricate hierarchical structures with dimensions 100 nm several microns. Using machine learning (ML) based Gaussian process regression algorithm, successfully identified certain topographical signals that either repress (pro-M2) or stimulate (pro-M1) polarization. upscaling nanotopographies further examination has shown mechanisms such as cytoskeletal remodeling ROCK-dependent epigenetic activation mechanotransduction pathways regulating fate. Thus, have also platform combining advanced DLIL nanofabrication techniques, HTS, ML-driven prediction nanobio interactions, pathway evaluation. short, our technology not only improves investigate understand nanotopography-regulated but holds great guiding design nanostructured coatings therapeutic biomaterials

Язык: Английский

---

Recent Progress in Organic Electrochemical Transistor-Structured Biosensors

Biosensors, Год журнала: 2024, Номер 14(7), С. 330 - 330

Опубликована: Июль 4, 2024

The continued advancement of organic electronic technology will establish electrochemical transistors as pivotal instruments in the field biological detection. Here, we present a comprehensive review state-of-the-art and advancements use biosensors. This provides an in-depth analysis diverse modification materials, methods, mechanisms utilized transistor-structured biosensors (OETBs) for selective detection wide range target analyte encompassing electroactive species, electro-inactive cancer cells. Recent advances OETBs sensing systems wearable implantable applications are also briefly introduced. Finally, challenges opportunities discussed.

Язык: Английский

---

Is deep brain imaging on the brink of transformation with a bioluminescence molecule?

BMEMat, Год журнала: 2024, Номер 2(3)

Опубликована: Авг. 16, 2024

Abstract Cephalofurimazine (CFz), when paired with Antares luciferase, shows superior blood‐brain barrier permeability and enhanced imaging depth clarity for deep brain imaging. This bioluminescence provides a less invasive method real‐time monitoring of activity, the potential to advance targeted therapies deepen our understanding functions. Further molecular engineering localized delivery can reduce toxicity CFz enhance its efficacy clinical

Язык: Английский

---

Analyzing the Impact of Binaural Beats on Anxiety Levels by a New Method Based on Denoised Harmonic Subtraction and Transient Temporal Feature Extraction

Bioengineering, Год журнала: 2024, Номер 11(12), С. 1251 - 1251

Опубликована: Дек. 10, 2024

Anxiety is a widespread mental health issue, and binaural beats have been explored as potential non-invasive treatment. EEG data reveal changes in neural oscillation connectivity linked to anxiety reduction; however, harmonics introduced during signal acquisition processing often distort these findings. Existing methods struggle effectively reduce capture the fine-grained temporal dynamics of signals, leading inaccurate feature extraction. Hence, novel Denoised Harmonic Subtraction Transient Temporal Feature Extraction proposed improve analysis impact on levels. Initially, Wiener Fused Convo Filter spatial features eliminate linear noise signals. Next, an Intrinsic Network employed, utilizing Attentive Weighted Least Mean Square (AW-LMS) algorithm nonlinear summation resonant coupling effects, eliminating misinterpretation brain rhythms. To address challenge dynamics, Embedded Transfo XL Recurrent detect extract relevant parameters associated with transient events data. Finally, undergo harmonic reduction extraction before classification cross-correlated Markov Deep Q-Network (DQN). This facilitates level into normal, mild, moderate, severe categories. The model demonstrated high accuracy 95.6%, precision 90%, sensitivity 93.2%, specificity 96% classifying levels, outperforming previous models. integrated approach enhances processing, enabling reliable offering valuable insights for therapeutic interventions.

Язык: Английский

---

The Effect of Processing Techniques on the Classification Accuracy of Brain-Computer Interface Systems

Brain Sciences, Год журнала: 2024, Номер 14(12), С. 1272 - 1272

Опубликована: Дек. 18, 2024

Background/Objectives: Accurately classifying Electroencephalography (EEG) signals is essential for the effective operation of Brain-Computer Interfaces (BCI), which needed reliable neurorehabilitation applications. However, many factors in processing pipeline can influence classification performance. The objective this study to assess effects different steps on accuracy EEG-based BCI systems. Methods: This explores impact various techniques and stages, including FASTER algorithm artifact rejection (AR), frequency filtering, transfer learning, cropped training. Physionet dataset, consisting four motor imagery classes, was used as input due its relatively large number subjects. raw EEG tested with EEGNet Shallow ConvNet. To examine adding a spatial dimension data, we also Multi-branch Conv3D Net developed two new models, Conv2D Net. Results: Our analysis showed that be affected by at every stage. Applying AR method, instance, either enhance or degrade performance, depending subject specific network architecture. Transfer learning improving performance all networks both artifact-rejected data. improvement data less pronounced compared unfiltered resulting reduced precision. For best classifier achieved 46.1% increased 63.5% learning. In filtered case, rose from 45.5% only 55.9% when applied. An unexpected outcome regarding filtering observed: demonstrated better focusing lower-frequency components. Higher ranges were more discriminative ConvNet, but training Conclusions: findings highlight complex interaction between neural emphasizing necessity customized approaches tailored subjects architectures.

Язык: Английский

---