A Classification Method for Multichannel MI-EEG Signal with FPCA and DNN DOI Open Access

Yunhui Hou,

Na Shen, Yubin Lin

et al.

Journal of Physics Conference Series, Journal Year: 2024, Volume and Issue: 2891(11), P. 112014 - 112014

Published: Dec. 1, 2024

Abstract A new accurate identification method has been proposed to address the lack of interpretability in current deep learning-based feature extraction methods for motor imagery electroencephalogram (MI-EEG) signals. This combines functional principal component analysis (FPCA) and neural networks (DNN) four classifications MI-EEG The process involves preprocessing acquired signals obtaining power spectral density (PSD) versus frequency curves alpha band multiple channels samples through FIR filtering. All PSD-frequency are then functionally smoothed according theory data (FDA). Feature parameters derived using FPCA, all normalized. Training selected randomly clustering training with DNNs. Category prediction is carried out on test classification samples. applied 4×120 four-categorized samples, each from six obtained Enobio test, a wireless EEG system Spain Neuroelectrics, involving left hand, right foot, foot at sampling rate 500Hz. 80% were used training, remaining 20% testing. accuracy ranged 84.3% 91.66%. While this multivariate parameter clear mathematical physical significance, it does demand high

Language: Английский

Recent applications of EEG-based brain-computer-interface in the medical field DOI Creative Commons
Xiuyun Liu, Wenlong Wang, Miao Liu

et al.

Military Medical Research, Journal Year: 2025, Volume and Issue: 12(1)

Published: March 24, 2025

Abstract Brain-computer interfaces (BCIs) represent an emerging technology that facilitates direct communication between the brain and external devices. In recent years, numerous review articles have explored various aspects of BCIs, including their fundamental principles, technical advancements, applications in specific domains. However, these reviews often focus on signal processing, hardware development, or limited such as motor rehabilitation communication. This paper aims to offer a comprehensive electroencephalogram (EEG)-based BCI medical field across 8 critical areas, encompassing rehabilitation, daily communication, epilepsy, cerebral resuscitation, sleep, neurodegenerative diseases, anesthesiology, emotion recognition. Moreover, current challenges future trends BCIs were also discussed, personal privacy ethical concerns, network security vulnerabilities, safety issues, biocompatibility.

Language: Английский

Citations

2
Holographic convolutional attention neural network for motor imagery decoding based on EEG temporal–spatial frequency features DOI
Qingsong Ai,

Yuang Liu,

Quan Liu

et al.

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 104, P. 107526 - 107526

Published: Jan. 29, 2025

Language: Английский

Citations

1
A Novel Approach to Motor Imagery EEG Signal Transformation and Classification Using Stockwell Transform and Deep Learning Models DOI Open Access
Çağatay Murat Yılmaz

Karadeniz Fen Bilimleri Dergisi, Journal Year: 2025, Volume and Issue: 15(1), P. 152 - 170

Published: March 15, 2025

Motor imagery (MI) classification using EEG signals has gained popularity, playing an essential role in developing technologies such as brain-computer interfaces (BCIs). This paper proposes novel approaches the Stockwell transform (S-transform) to encode into images time-frequency space and classify them by feeding pre-trained Inception-ResNet-V2, AlexNet, SqueezeNet CNNs. High subject-to-subject session-to-session signal variability hinder recognition of MI tasks. Most literature studied within-subject performance. study conducted experiments a leave-one-subject-out cross-validation strategy, investigated inter-subject variation's effect contributed evaluating model's performance generalization ability. At same time, different sessions presence or absence feedback were assessed, results analyzed. The are encouraging, considering difficulty classifying differences. For cue-based paradigm non-feedback signals, between 62.1-80.8%; for with smiley feedback, 57.1-96.3%; without 56.8-91.4%. These findings highlight potential combining S-transform CNNs, offering valuable insights EEG-based BCI applications.

Language: Английский

Citations

1
Generative Diffusion-Based Task Incremental Learning Method for Decoding Motor Imagery EEG DOI Creative Commons
Yufei Yang, Mingai Li, Jianhang Liu

et al.

Brain Sciences, Journal Year: 2025, Volume and Issue: 15(2), P. 98 - 98

Published: Jan. 21, 2025

Background/Objectives: Motor neurorehabilitation can be realized by gradually learning diverse motor imagery (MI) tasks. EEG-based brain-computer interfaces (BCIs) provide an effective solution. Nevertheless, existing MI decoding methods cannot balance plasticity for unseen tasks and stability old This paper proposes a generative diffusion-based task Incremental Learning (IL) method called GD-TIL. Methods: First, data augmentation is employed to increase diversity segmenting recombining EEG signals. Second, capture temporal-spatial features (TSFs) from different temporal resolutions, multi-scale feature extractor (MTSFE) developed via integrating multiscale convolutions, dual-branch pooling operation, multiple multi-head self-attention mechanisms, dynamic convolutional encoder. The proposed self-supervised generalization (SSTG) mechanism introduces regularization constraint guide MTSFE unified classifier updating, which combines labels semantic similarity between the with original views enhance model generalizability In IL phase, prototype-guided replay module (PGGR) used generate tasks’ TSFs training lightweight diffusion based on prototype label of each task. Furthermore, generated TSF merged new fine-tune encoder update PGGR. Finally, GD-TIL evaluated self-collected ADL-MI dataset two pairs public four Results: continuous accuracy reaches 80.20% 81.32%, respectively. experimental results exhibit excellent GD-TIL, even beating state-of-the-art methods. Conclusions: Our work illustrates potential MI-based BCI AI neurorehabilitation.

Language: Английский

Citations

0
A session-incremental broad learning system for motor imagery EEG classification DOI
Yufei Yang, Mingai Li, Hanlin Liu

et al.

Biomedical Signal Processing and Control, Journal Year: 2024, Volume and Issue: 97, P. 106717 - 106717

Published: Aug. 14, 2024

Language: Английский

Citations

3
An effective classification approach for EEG-based motor imagery tasks combined with attention mechanisms DOI
Jixiang Li, Wuxiang Shi, Yurong Li

et al.

Cognitive Neurodynamics, Journal Year: 2024, Volume and Issue: 18(5), P. 2689 - 2707

Published: May 3, 2024

Language: Английский

Citations

2
Decoding lower-limb kinematic parameters during pedaling tasks using deep learning approaches and EEG DOI
Cristian Felipe Blanco-Díaz, Cristian David Guerrero-Méndez, Rafhael M. Andrade

et al.

Medical & Biological Engineering & Computing, Journal Year: 2024, Volume and Issue: 62(12), P. 3763 - 3779

Published: July 19, 2024

Language: Английский

Citations

2
EEG-VTTCNet: A loss joint training model based on the vision transformer and the temporal convolution network for EEG-based motor imagery classification DOI
Xingbin Shi, Baojiang Li, Wenlong Wang

et al.

Neuroscience, Journal Year: 2024, Volume and Issue: 556, P. 42 - 51

Published: Aug. 3, 2024

Language: Английский

Citations

2
An adaptive session-incremental broad learning system for continuous motor imagery EEG classification DOI
Yufei Yang, Mingai Li, Linlin Wang

et al.

Medical & Biological Engineering & Computing, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 29, 2024

Language: Английский

Citations

1
Antidepressant Treatment Response Prediction with Early Assessment of Functional Near-Infrared Spectroscopy and Micro-RNA DOI Creative Commons
Lok Hua Lee, Cyrus S. H. Ho, Yee Ling Chan

et al.

IEEE Journal of Translational Engineering in Health and Medicine, Journal Year: 2024, Volume and Issue: 13, P. 9 - 22

Published: Nov. 26, 2024

While functional near-infrared spectroscopy (fNIRS) had previously been suggested for major depressive disorder (MDD) diagnosis, the clinical application to predict antidepressant treatment response (ATR) is still unclear. To address this, aim of current study investigate MDD ATR at three levels using fNIRS and micro-ribonucleic acids (miRNAs). Our proposed algorithm includes a custom inter-subject variability reduction based on principal component analysis (PCA). The components extracted features are first identified non-responders' group. few that sum up 99% explained variance discarded minimize while remaining projection vectors applied all groups (24 non-responders, 15 partial-responders, 13 responders) obtain their relative projections in feature space. entire achieved better performance through radial basis function (RBF) support vector machine (SVM), with 82.70% accuracy, 78.44% sensitivity, 86.15% precision, 91.02% specificity, respectively, when compared conventional learning approaches combine clinical, sociodemographic genetic information as predictor. suggests prediction can be improved multiple sources, provided properly addressed, an effective tool decision system prediction. Clinical Translational Impact Statement-The fusion neuroimaging miRNA profiles significantly enhances accuracy ATR. minimally required also make personalized medicine more practical realizable.

Language: Английский

Citations

0