Msrtnet: Multi-Scale Spatial Residual Network Based on Time-Domain Transformer DOI
Xin Gao, Dingguo Zhang, Xiaolong Wu

и другие.

Опубликована: Янв. 1, 2024

Язык: Английский

Multichannel convolutional transformer for detecting mental disorders using electroancephalogrpahy records DOI Creative Commons
Mamadou Dia,

Ghazaleh Khodabandelou,

Syed Muhammad Anwar

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Май 2, 2025

Mental disorders represent a critical global health challenge that affects millions around the world and significantly disrupts daily life. Early accurate detection is paramount for timely intervention, which can lead to improved treatment outcomes. Electroencephalography (EEG) provides non-invasive means observing brain activity, making it useful tool detecting potential mental disorders. Recently, deep learning techniques have gained prominence their ability analyze complex datasets, such as electroencephalography recordings. In this study, we introduce novel deep-learning architecture classification of post-traumatic stress disorder, depression, or anxiety, using data. Our proposed model, multichannel convolutional transformer, integrates strengths both neural networks transformers. Before feeding model low-level features, input pre-processed common spatial pattern filter, signal space projection wavelet denoising filter. Then EEG signals are transformed continuous transform obtain time-frequency representation. The layers tokenize by our pre-processing pipeline, while Transformer encoder effectively captures long-range temporal dependencies across sequences. This specifically tailored process data has been preprocessed transform, technique representation, thereby enhancing extraction relevant features classification. We evaluated performance on three datasets: Psychiatric Dataset, MODMA dataset, Psychological Assessment dataset. achieved accuracies 87.40% 89.84% 92.28% approach outperforms every concurrent approaches datasets used, without showing any sign over-fitting. These results underscore in delivering reliable disorder through analysis, paving way advancements early diagnosis strategies.

Язык: Английский

Процитировано

0

EEG-based visual stimuli classification via reusable LSTM DOI
Yaling Deng, Shuo Ding,

Wenyi Li

и другие.

Biomedical Signal Processing and Control, Год журнала: 2023, Номер 82, С. 104588 - 104588

Опубликована: Янв. 13, 2023

Язык: Английский

Процитировано

7

TDLNet: Transfer Data Learning Network for Cross-Subject Classification Based on Multiclass Upper Limb Motor Imagery EEG DOI Creative Commons
Jingfeng Bi, Ming Chu

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Год журнала: 2023, Номер 31, С. 3958 - 3967

Опубликована: Янв. 1, 2023

The limited number of brain-computer interface based on motor imagery (MI-BCI) instruction sets for different movements single limbs makes it difficult to meet practical application requirements. Therefore, designing a single-limb, multi-category (MI) paradigm and effectively decoding is one the important research directions in future development MI-BCI. Furthermore, major challenges MI-BCI difficulty classifying brain activity across individuals. In this article, transfer data learning network (TDLNet) proposed achieve cross-subject intention recognition multiclass upper limb imagery. TDLNet, Transfer Data Module (TDM) used process electroencephalogram (EEG) signals groups then fuse channel features through two one-dimensional convolutions. Residual Attention Mechanism (RAMM) assigns weights each EEG signal dynamically focuses channels most relevant specific task. Additionally, feature visualization algorithm occlusion frequency qualitatively analyze TDLNet. experimental results show that TDLNet achieves best classification datasets compared CNN-based reference methods method. 6-class scenario, obtained an accuracy 65%±0.05 UML6 dataset 63%±0.06 GRAZ dataset. demonstrate framework can produce distinct classifier patterns multiple categories frequencies. ULM6 available at https://dx.doi.org/10.21227/8qw6-f578.

Язык: Английский

Процитировано

7

Flexible coding scheme for robotic arm control driven by motor imagery decoding DOI
Qingsong Ai,

Mengyuan Zhao,

Kun Chen

и другие.

Journal of Neural Engineering, Год журнала: 2022, Номер 19(5), С. 056008 - 056008

Опубликована: Июль 27, 2022

Objective. Brain computer interface (BCI) technology is an innovative way of information exchange, which can effectively convert physiological signals into control instructions machines. Due to its spontaneity and device independence, the motor imagery (MI) electroencephalography (EEG) signal used as a common BCI source achieve direct external devices. Several online MI EEG-based systems have shown potential for rehabilitation. However, generalization ability current classification model tasks still limited real-time prototype far from widespread in practice.Approach. To solve these problems, this paper proposes optimized neural network architecture based on our previous work. Firstly, artifact components MI-EEG are removed by using threshold function related removal evaluation index, then data augmented empirical mode decomposition (EMD) algorithm. Furthermore, ensemble learning (EL) method fine-tuning strategy transfer (TL) optimize model. Finally, combined with flexible binary encoding strategy, EEG recognition results mapped commands robotic arm, realizes multiple degrees freedom arm.Main results. The show that EMD has obvious amount enhancement effect small dataset, EL TL improve intra-subject inter-subject performance, respectively. use coding expansion instructions, i.e. four kinds complete 7 arm.Significance. Our work not only improves accuracy subject generality while also extending instruction set.

Язык: Английский

Процитировано

11

Graph neural network based on brain inspired forward-forward mechanism for motor imagery classification in brain-computer interfaces DOI Creative Commons
Qiwei Xue, Yuntao Song, Huapeng Wu

и другие.

Frontiers in Neuroscience, Год журнала: 2024, Номер 18

Опубликована: Март 28, 2024

Introduction Within the development of brain-computer interface (BCI) systems, it is crucial to consider impact brain network dynamics and neural signal transmission mechanisms on electroencephalogram-based motor imagery (MI-EEG) tasks. However, conventional deep learning (DL) methods cannot reflect topological relationship among electrodes, thereby hindering effective decoding activity. Methods Inspired by concept neuronal forward-forward (F-F) mechanism, a novel DL framework based Graph Neural Network combined mechanism (F-FGCN) presented. F-FGCN aims enhance EEG performance applying functional relationships propagation mechanism. The fusion process involves converting multi-channel into sequence signals constructing grounded Pearson correlation coeffcient, effectively representing associations between channels. Our model initially pre-trains Convolutional (GCN), fine-tunes output layer obtain feature vector. Moreover, F-F used for advanced extraction classification. Results discussion Achievement assessed PhysioNet dataset four-class categorization, compared with various classical state-of-the-art models. learned features substantially amplify downstream classifiers, achieving highest accuracy 96.11% 82.37% at subject group levels, respectively. Experimental results affirm potency FFGCN in enhancing performance, thus paving way BCI applications.

Язык: Английский

Процитировано

2

Enhancing motor imagery EEG signal decoding through machine learning: A systematic review of recent progress DOI
Ibrahim A. Hameed, Danish M. Khan,

Samiha Ahmed

и другие.

Computers in Biology and Medicine, Год журнала: 2024, Номер 185, С. 109534 - 109534

Опубликована: Дек. 12, 2024

Язык: Английский

Процитировано

2

EEG temporal information-based 1-D convolutional neural network for motor imagery classification DOI
Chaoqin Chu,

Qinkun Xiao,

Leran Chang

и другие.

Multimedia Tools and Applications, Год журнала: 2023, Номер 82(29), С. 45747 - 45767

Опубликована: Авг. 22, 2023

Язык: Английский

Процитировано

5

Predicting components of pulpwood feedstock for different physical forms and tree species using NIR spectroscopy and transfer learning DOI
Zheyu Zhang, Hao Zhong, Yaoxiang Li

и другие.

Cellulose, Год журнала: 2023, Номер unknown

Опубликована: Ноя. 25, 2023

Язык: Английский

Процитировано

5

MOTOR IMAGERY SIGNAL CLASSIFICATION FOR BRAIN–COMPUTER INTERFACE USING RideNN WITH HOLO-ENTROPY FEATURES DOI
Megha M. Wankhade, S. S. Chorage

Biomedical Engineering Applications Basis and Communications, Год журнала: 2024, Номер 36(04)

Опубликована: Июль 3, 2024

The brain–computer interface (BCI) database’s motor assessment depends heavily on the imagery (MI) signal classification. By examining multiple patterns of different creative tasks in electroencephalogram (EEG) signals, intention humans is translated into computer-based commands MI-based signals BCI data. Nevertheless, low accuracy and efficiency are issues with MI–EEG signals’ classification because signal-to-noise ratio, huge individual differences, overall volatility, complexity signal. To overcome these problems, this research proposes a rider optimization algorithm-based neural network (ROA-based NN) to classify MI effectively. Pre-processing done after collecting dataset raw EEG signals. suitable electrodes, such as C3, C4, Cz, subsequently chosen from Using holo-entropy-based WPD feature extractor, tunable Q-factor wavelet transform (T-QWT), common spatial model, pertinent features extracted electrodes. developed examines electrode structure’s association. As result, most diverse removed electrodes before being input proposed RideNN classifier, where ride algorithm optimizes performance correctly predicts classes output that have been analyzed. classifier recognizes more accurately processes data tackles noise incomplete Utilizing parameters accuracy, sensitivity, specificity, results evaluated. PROA-based combined obtain maximum 92.24%, sensitivity 92.26%, specificity 92.14% for competition-IV 2a database. qPROA-based 92.11%, 91.98%, 92.35% 2b

Язык: Английский

Процитировано

1

Cross-subject emotion EEG signal recognition based on source microstate analysis DOI Creative Commons
Lei Zhang, Di Xiao,

Xiaojing Guo

и другие.

Frontiers in Neuroscience, Год журнала: 2023, Номер 17

Опубликована: Ноя. 28, 2023

Electroencephalogram (EEG) signals are very weak and have low spatial resolution, which has led to less satisfactory accuracy in cross-subject EEG-based emotion classification studies. Microstate analyses of EEG sources can be performed determine the important spatiotemporal characteristics signals. Such used cluster rapidly changing into multiple brain prototype topographies, fully utilizing information contained providing a neural representation for emotional dynamics. To better utilize signals, source localization analysis on was first conducted. Then, microstate source-reconstructed is conducted extract features data. We participant data from odor-video physiological signal database (OVPD-II) dataset. The experimental results show that feature topologies different participants under same exhibited high degree correlation, proven by topographic maps comparison two-dimensional visualization differential entropy (DE) power spectral density (PSD). represent more abstract robust. extracted were then with style transfer mapping method domain target support vector machines (SVMs) convolutional networks (CNNs) recognition. accuracies SVMs 84.90 ± 8.24% 87.43 7.54%, 7.19 6.95% higher than those obtained PSD 0.51 1.79% DE features. In CNN, average 86.44 91.49%, 7.71 19.41% 2.7 11.76%

Язык: Английский

Процитировано

3