Implementation of artificial intelligence and machine learning-based methods in brain–computer interaction

Computers in Biology and Medicine, Journal Year: 2023, Volume and Issue: 163, P. 107135 - 107135

Published: June 8, 2023

Brain–computer interfaces are used for direct two-way communication between the human brain and computer. Brain signals contain valuable information about mental state activity of examined subject. However, due to their non-stationarity susceptibility various types interference, processing, analysis interpretation challenging. For these reasons, research in field brain–computer is focused on implementation artificial intelligence, especially five main areas: calibration, noise suppression, communication, condition estimation, motor imagery. The use algorithms based intelligence machine learning has proven be very promising application domains, ability predict learn from previous experience. Therefore, within medical technologies can contribute more accurate subjects, alleviate consequences serious diseases or improve quality life disabled patients.

Language: Английский

---

Cross-Subject Emotion Recognition Brain–Computer Interface Based on fNIRS and DBJNet

Cyborg and Bionic Systems, Journal Year: 2023, Volume and Issue: 4

Published: Jan. 1, 2023

Functional near-infrared spectroscopy (fNIRS) is a noninvasive brain imaging technique that has gradually been applied in emotion recognition research due to its advantages of high spatial resolution, real time, and convenience. However, the current on based fNIRS mainly limited within-subject, there lack related work across subjects. Therefore, this paper, we designed an evoking experiment with videos as stimuli constructed database. On basis, deep learning technology was introduced for first dual-branch joint network (DBJNet) constructed, creating ability generalize model new participants. The decoding performance obtained by proposed shows can effectively distinguish positive versus neutral negative emotions (accuracy 74.8%, F1 score 72.9%), 2-category task distinguishing 89.5%, 88.3%), 91.7%, 91.1%) proved powerful decode emotions. Furthermore, results ablation study structure demonstrate convolutional neural branch statistical achieve highest performance. paper expected facilitate development affective brain-computer interface.

Language: Английский

---

Interpretable deep learning model for major depressive disorder assessment based on functional near-infrared spectroscopy

Asian Journal of Psychiatry, Journal Year: 2024, Volume and Issue: 92, P. 103901 - 103901

Published: Jan. 3, 2024

Language: Английский

---

Brain-computer interface paradigms and neural coding

Frontiers in Neuroscience, Journal Year: 2024, Volume and Issue: 17

Published: Jan. 15, 2024

Brain signal patterns generated in the central nervous system of brain-computer interface (BCI) users are closely related to BCI paradigms and neural coding. In systems, coding critical elements for research. However, so far there have been few references that clearly systematically elaborated on definition design principles paradigm as well modeling Therefore, these contents expounded existing main introduced review. Finally, challenges future research directions were discussed, including user-centered evaluation coding, revolutionizing traditional paradigms, breaking through techniques collecting brain signals combining technology with advanced AI improve decoding performance. It is expected review will inspire innovative development

Language: Английский

---

EF-Net: Mental State Recognition by Analyzing Multimodal EEG-fNIRS via CNN

Sensors, Journal Year: 2024, Volume and Issue: 24(6), P. 1889 - 1889

Published: March 15, 2024

Analysis of brain signals is essential to the study mental states and various neurological conditions. The two most prevalent noninvasive for measuring activities are electroencephalography (EEG) functional near-infrared spectroscopy (fNIRS). EEG, characterized by its higher sampling frequency, captures more temporal features, while fNIRS, with a greater number channels, provides richer spatial information. Although few previous studies have explored use multimodal deep-learning models analyze activity both EEG subject-independent training–testing split analysis remains underexplored. results setting directly show model’s ability on unseen subjects, which crucial real-world applications. In this paper, we introduce EF-Net, new CNN-based model. We evaluate EF-Net an EEG-fNIRS word generation (WG) dataset state recognition task, primarily focusing setting. For completeness, report in subject-dependent subject-semidependent settings as well. compare our model five baseline approaches, including three traditional machine learning methods deep methods. demonstrates superior performance accuracy F1 score, surpassing these baselines. Our achieves scores 99.36%, 98.31%, 65.05% subject-dependent, subject-semidependent, settings, respectively, best 1.83%, 4.34%, 2.13% These highlight EF-Net’s capability effectively learn interpret across different subjects.

Language: Английский

---

Effects of tactile feedback in post-stroke hand rehabilitation on functional connectivity and cortical activation: an fNIRS study

Biomedical Optics Express, Journal Year: 2025, Volume and Issue: 16(2), P. 643 - 643

Published: Jan. 6, 2025

Stroke-induced hand motor impairments have a significant impact on the daily lives of patients. Motor rehabilitation with tactile feedback (TF) shows promise as an effective intervention; however, its neural mechanisms are still not fully understood. The main objective this study was to examine effect brain functional responses during single movement session in post-stroke patients, using near-infrared spectroscopy (fNIRS). changes oxy- and deoxy-hemoglobin concentrations were recorded from bilateral prefrontal, motor, occipital areas 13 patients subacute recovery phase 15 healthy controls hand-grasping task TF no-TF. cortical activation responses, connectivity, network properties calculated explore specific response two grasping tasks. results showed that exhibited increased hemodynamic cortex tasks TF. However, prefrontal cortex, left sensorimotor right premotor area significantly lower compared (p < 0.05). Additionally, poorer overall function, reductions both clustering coefficient = 0.0016), reflecting local information transfer efficiency, transitivity 0.0053), representing global integration. A positive correlation observed between grip strength metrics (r 0.592, p 0.033), well 0.590, 0.034) indicating greater associated reduced transmission efficiency. These findings indicated can modulate activity learning integration, providing evidence for potential valuable tool stroke rehabilitation.

Language: Английский

---

Comparing Multi-Dimensional fNIRS Features Using Bayesian Optimization-Based Neural Networks for Mild Cognitive Impairment (MCI) Detection

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2023, Volume and Issue: 31, P. 1019 - 1029

Published: Jan. 1, 2023

The diagnosis of mild cognitive impairment (MCI), a prodromal stage Alzheimer's disease (AD), is essential for initiating timely treatment to delay the onset AD. Previous studies have shown potential functional near-infrared spectroscopy (fNIRS) diagnosing MCI. However, preprocessing fNIRS measurements requires extensive experience identify poor-quality segments. Moreover, few explored how proper multi-dimensional features influence classification results disease. Thus, this study outlined streamlined method process and compared with neural networks in order explore temporal spatial factors affect MCI normality. More specifically, proposed using Bayesian optimization-based auto hyperparameter tuning evaluate 1D channel-wise, 2D spatial, 3D spatiotemporal detecting patients. highest test accuracies 70.83%, 76.92%, 80.77% were achieved 1D, 2D, features, respectively. Through comparisons, time-point oxyhemoglobin feature was proven be more promising by an dataset 127 participants. Furthermore, presented approach data processing, designed models required no manual tuning, which promoted general utilization modality network-based detect

Language: Английский

---

Temporal Convolutional Network-Enhanced Real-Time Implicit Emotion Recognition with an Innovative Wearable fNIRS-EEG Dual-Modal System

Electronics, Journal Year: 2024, Volume and Issue: 13(7), P. 1310 - 1310

Published: March 31, 2024

Emotion recognition remains an intricate task at the crossroads of psychology and artificial intelligence, necessitating real-time, accurate discernment implicit emotional states. Here, we introduce a pioneering wearable dual-modal device, synergizing functional near-infrared spectroscopy (fNIRS) electroencephalography (EEG) to meet this demand. The first-of-its-kind fNIRS-EEG ensemble exploits temporal convolutional network (TC-ResNet) that takes 24 fNIRS 16 EEG channels as input for extraction features. Our system has many advantages including its portability, battery efficiency, wireless capabilities, scalable architecture. It offers real-time visual interface observation cerebral electrical hemodynamic changes, tailored variety real-world scenarios. approach is comprehensive detection strategy, with new designs in architecture deployment improvement signal processing interpretation. We examine interplay emotions physiological responses elucidate cognitive processes emotion regulation. An extensive evaluation 30 subjects under four induction protocols demonstrates our bimodal system’s excellence detecting emotions, impressive classification accuracy 99.81% ability reveal interconnection between signals. Compared latest unimodal identification methods, shows significant gains 0.24% 8.37% fNIRS. Moreover, proposed TC-ResNet-driven fusion technique outperforms conventional EEG-fNIRS improving from 0.7% 32.98%. This research presents groundbreaking advancement affective computing combines biological engineering intelligence. integrated solution facilitates nuanced responsive intelligence practical applications, far-reaching impacts on personalized healthcare, education, human–computer interaction paradigms.

Language: Английский

---

Diagnostic machine learning applications on clinical populations using functional near infrared spectroscopy: a review

Reviews in the Neurosciences, Journal Year: 2024, Volume and Issue: 35(4), P. 421 - 449

Published: Feb. 3, 2024

Abstract Functional near-infrared spectroscopy (fNIRS) and its interaction with machine learning (ML) is a popular research topic for the diagnostic classification of clinical disorders due to lack robust objective biomarkers. This review provides an overview on psychiatric diseases by using fNIRS ML. Article search was carried out 45 studies were evaluated considering their sample sizes, used features, ML methodology, reported accuracy. To our best knowledge, this first that reports applications fNIRS. We found there has been increasing trend perform fNIRS-based biomarker since 2010. The most studied populations are schizophrenia ( n = 12), attention deficit hyperactivity disorder 7), autism spectrum 6) populations. There significant negative correlation between size (>21) accuracy values. Support vector (SVM) deep (DL) approaches classifier (SVM 20) (DL 10). Eight these recruited number participants more than 100 classification. Concentration changes in oxy-hemoglobin (ΔHbO) based features concentration deoxy-hemoglobin (ΔHb) ones ΔHbO-based mean ΔHbO 11) functional connections 11). Using data might be promising approach reveal specific biomarkers

Language: Английский

---

Application of artificial intelligence in cognitive load analysis using functional near-infrared spectroscopy: A systematic review

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 249, P. 123717 - 123717

Published: March 22, 2024

Cognitive load theory suggests that overloading of working memory may negatively affect the performance human in cognitively demanding tasks. Evaluation cognitive is a difficult task; it often assessed through feedback and evaluation from experts. classification based on Functional Near-InfraRed Spectroscopy (fNIRS) now one key research areas recent years, due to its resistance artefacts, cost-effectiveness, portability. To make fNIRS more practical various applications, necessary develop robust algorithms can automatically classify signals less reliant trained signals. Many analytical tools used sciences have Deep Learning (DL) modalities uncover relevant information for mental workload classification. This review investigates questions design overall effectiveness DL as well characteristics. We identified 38 studies published between 2011 2022, specifically proposed Machine (ML) models classifying using data obtained devices. Those were analyzed type feature selection methods, input, model architectures. Most existing are ML algorithms, which follow signal filtration hand-crafted features. It observed hybrid architectures integrate convolution LSTM operators performed significantly better comparison with other models. However, especially not been extensively investigated captured by The current trends challenges highlighted provide directions development pertaining research.

Language: Английский

---