Wyniki wyszukiwania - BazTech

1

Decoding motor imagery based on dipole feature imaging and a hybrid CNN with embedded squeeze-and-excitation block

Wang Linlin, Li Mingai

Biocybernetics and Biomedical Engineering

|

2023

|

Vol. 43, no. 4

751--762

EN

Motor imagery (MI) decoding is the core of an intelligent rehabilitation system in brain computer interface, and it has a potential advantage by using source signals, which have higher spatial resolution and the same time resolution compared to scalp electroencephalography (EEG). However, how to delve and utilize the personalized frequency characteristic of dipoles for improving decoding performance has not been paid sufficient attention. In this paper, a novel dipole feature imaging (DFI) and a hybrid convolutional neural network (HCNN) with an embedded squeeze-and-excitation block (SEB), denoted as DFI-HCNN, are proposed for decoding MI tasks. EEG source imaging technique is used for brain source estimation, and each sub-band spectrum powers of all dipoles are calculated through frequency analysis and band division. Then, the 3D space information of dipoles is retrieved, and by using azimuthal equidistant projection algorithm it is transformed to a 2D plane, which is combined with nearest neighbor interpolation to generate multi sub-band dipole feature images. Furthermore, a HCNN is designed and applied to the ensemble of sub-band dipole feature images, from which the importance of sub-bands is acquired to adjust the corresponding attentions adaptively by SEB. Ten-fold cross-validation experiments on two public datasets achieve the comparatively higher decoding accuracies of 84.23% and 92.62%, respectively. The experiment results show that DFI is an effective feature representation, and HCNN with an embedded SEB can enhance the useful frequency information of dipoles for improving MI decoding.

2

Analysis of the application of brain-computer interfaces of a selected paradigm in everyday life

Mróz Katarzyna, Plechawska-Wójcik Małgorzata

Journal of Computer Sciences Institute

|

2022

|

Vol. 23

118--122

EN

The main objective of this paper is to carry out a research on the analysis of the use of brain-computer interface in everyday life. The article presents the method of recording brain activity, electroencephalography, which was used in the study. The brain activity used in the brain-computer interface and the general principle of brain-computer interface design are also described. The performed study allowed to develop an analysis of the obtained results in the matter of evaluating the usability of brain-computer interfaces using motor imagery. As a result of the process of analyzing the results obtained during the research, it was found that each subsequent experiment allowed for obtaining more favourable results than the previous one. The reason for this was the use of an additional training session for the next test person. In the final stage, it was possible to evaluate the usability of the brain-computer interface in everyday life

PL

Głównym celem artykułu jest przeprowadzenie badania nad analizą wykorzystania interfejsu mózg-komputer w życiu codziennym. W artykule przedstawiono metodę rejestrowania aktywności mózgu, elektroencefalografię, która została wykorzystana w badaniu. Opisano również aktywność mózgu wykorzystywaną w interfejsie mózg-komputer oraz ogólną zasadę projektowania interfejsu mózg-komputer. Przeprowadzone badanie pozwoliło na opracowanie analizy uzyskanych wyników w zakresie oceny użyteczności interfejsów mózg-komputer z wykorzystaniem obrazowania motorycznego. W wyniku procesu analizy wyników uzyskanych podczas przeprowadzania badań ustalono, iż każdy następnie zrealizowany eksperyment pozwalał na uzyskanie korzystniejszych wyników od poprzedniego. Powodem tego było zastosowanie dodatkowej sesji treningowej dla kolejnych badanych osób. W końcowym etapie można było ocenić przydatność interfejsu mózg-komputer w życiu codziennym

3

The quantitative application of channel importance in movement intention decoding

Wang Linlin, Li Mingai

Biocybernetics and Biomedical Engineering

|

2022

|

Vol. 42, no. 2

630--645

EN

The complex brain network consists of multiple collaborative regions, which can be activated to varying degrees by motor imagery (MI) and the induced electroencephalogram (EEG) recorded by an array of scalp electrodes is usually decoded for driving rehabilitation system. Either all channels or partially selected channels are equally applied to recognize movement intention, which may be incompatible with the individual differences of channels from different locations. In this paper, a channel importance based imaging method is proposed, denoted as CIBI. For each electrode of MI-EEG, the power over 8–30 Hz band is calculated from discrete Fourier spectrum and input to random forest algorithm (RF) to quantify its contribution, namely channel importance (CI); Then, CI is used for weighting the powers of α and β rhythms, which are interpolated to a 32 x 32 grid by using Clough-Tocher method respectively, generating two main band images with time-frequency-space information. In addition, a dual branch fusion convolutional neural network (DBFCNN) is developed to match with the characteristic of two MI images, realizing the extraction, fusion and classification of comprehensive features. Extensive experiments are conducted based on two public datasets with four classes of MI-EEG, the relatively higher average accuracies are obtained, and the improvements achieve 23:95% and 25:14% respectively when using channel importance, their statistical analysis are also performed by Kappa value, confusion matrix and receiver operating characteristic. Experiment results show that the personalized channel importance is helpful to enhance inter-class separability as well as the proposed method has the outstanding decoding ability for multiple MI tasks.

4

EEG_GENet: A feature-level graph embedding method for motor imagery classification based on EEG signals

Wang Huiyang, Yu Hua, Wang Haixian

Biocybernetics and Biomedical Engineering

|

2022

|

Vol. 42, no. 3

1023--1040

EN

In recent years, the success of deep learning has driven the development of motor imagery brain-computer interfaces (MI-BCIs) based on electroencephalography (EEG). However, unlike image or language data, motor imagery EEG signals are of multielectrodes with topology information. As a means of integrating graph topology information into feature maps, few studies studied motor imagery classification involving graph embeddings. To decode EEG signals more accurately, this paper proposes a feature-level graph embedding method and combines the method with EEGNet; this new network is called EEG_GENet. Specifically, time-domain features are obtained by convoluting raw EEG signals for each electrode. Then, the adjacent matrix, conceptualized as a graph filter, performs graph convolution and uses the time-domain features to embed the topology information. This process can also perform multi-order graph embeddings. In addition, the adjacency matrix in this paper can adapt to different brain network connectivities for different subjects. We evaluate the proposed method on two benchmark EEG datasets for motor imagery classification. Experimental results on the BCICIV-2a and High_Gamma datasets demonstrate that EEG_GENet achieves 79.57% and 96.02% classification accuracy, respectively. These results indicate that the proposed method is superior to state-of-the-art methods. In addition, various ablation experiments further verify the advantages of the feature-level graph embedding method. To conclude, the feature-level graph embedding method can improves the network’s ability to decode raw motor imagery EEG signals.

5

Review of solutions for the application of example of machine learning methods for Motor Imagery in correlation with Brain-Computer Interfaces

Paszkiel Szczepan, Rojek Ryszard, Lei Ningrong, Castro Maria Antonio

Przegląd Elektrotechniczny

|

2021

|

R. 97, nr 11

111--116

EN

Presently, numerous public databases presenting the collected EEG signals, including the ones in the scope of Motor Imagery (MI), are available. Simultaneously, machine-learning methods, which enable effective and fast discovering of information, also in the sets of biomedical data, are constantly being developed. In this paper, a set of 30 of some of the latest scientific publications from the years 2016-2021 has been analyzed. The analysis covered, among others: public data repositories in the form of EEG signals as input data; numbers and types of the analyzed tasks in the scope of MI in the above-mentioned databases; and Deep Learning (DL) architectures.

PL

Obecnie dostępne są liczne ogólnodostępne bazy danych prezentujące zebrane sygnały EEG, w tym z zakresu obrazowania motorycznego (MI). Jednocześnie stale rozwijane są metody uczenia maszynowego, które umożliwiają efektywne i szybkie odkrywanie informacji, także w zbiorach danych biomedycznych. W niniejszym artyule przeanalizowano zestaw 30 spośród najnowszych publikacji naukowych z lat 2016- 2021. Analizie poddano m.in.: publiczne repozytoria danych w postaci sygnałów EEG jako dane wejściowe; liczby i rodzaje analizowanych zadań z zakresu obrazowania motorycznego w ww. bazach; i architektury Deep Learning (DL).

6

Sequence-to-sequence deep neural network with spatio-spectro and temporal features for motor imagery classification

Rammy Sadaqat Ali, Abbas Waseem, Mahmood Syed Shahid, Riaz Haider, Rehman Haseeb Ur, Abideen Rao Zain Ul, Aqeel Muhammad, Zhang Wu

Biocybernetics and Biomedical Engineering

|

2021

|

Vol. 41, no. 1

97--110

EN

Electroencephalography (EEG) is a method of the brain–computer interface (BCI) that measures brain activities. EEG is a method of (non-)invasive recording ofthe electrical activity ofthe brain. This can be used to build BCIs. From the last decade, EEG has grasped researchers' attention to distinguish human activities. However, temporal information has rarely been retained to incorporate temporal information for multi-class (more than two classes) motor imagery classification. This research proposes a long-short-term-memory-based deep learning model to learn the hidden sequential patterns. Two types of features are used to feed the proposed model, including Fourier Transform Energy Maps (FTEMs) and Common Spatial Patterns (CSPs) filters. Multiple experiments have been conducted on a publicly available dataset. Extraction of spatial and spectro-temporal features using CSP filters and FTEM allow the sequence-tosequence based proposed model to learn the hidden sequential features. The proposed method is trained, evaluated, and optimized for a publicly available benchmark data set and resulted in 0.81 mean kappa value. Obtained results depict the model robustness for the artifacts and suitable for real-life applications with comparable classification accuracy. The code and findings will be available at https://github.com/waseemabbaas/Motor-Imagery-Classification.git.

7

Compact convolutional neural network (CNN) based on SincNet for end-to-end motor imagery decoding and analysis

Izzuddin Tarmizi Ahmad, Safri Norlaili Mat, Othman Mohd Afzan

Biocybernetics and Biomedical Engineering

|

2021

|

Vol. 41, no. 4

1629--1645

EN

In the field of human-computer interaction, the detection, extraction and classification of the electroencephalogram (EEG) spectral and spatial features are crucial towards developing a practical and robust non-invasive EEG-based brain-computer interface. Recently, due to the popularity of end-to-end deep learning, the applicability of algorithms such as convolutional neural networks (CNN) has been explored to achieve the mentioned tasks. This paper presents an improved and compact CNN algorithm for motor imagery decoding based on the adaptation of SincNet, which was initially developed for speaker recognition task from the raw audio input. Such adaptation allows for a compact end-to-end neural network with state-of-the-art (SOTA) performances and enables network interpretability for neurophysiological validation in cortical rhythms and spatial analysis. In order to validate the performance of proposed algorithms, two datasets were used; the first is the publicly available BCI Competition IV dataset 2a, which was often used as a benchmark in validating motor imagery classification algorithms, and the second is a dataset consists of primary data initially collected to study the difference between motor imagery and mental-task associated motor imagery BCI and was used to test the plausibility of the proposed algorithm in highlighting the differences in terms of cortical rhythms. Competitive decoding performance was achieved in both datasets in comparisons with SOTA CNN models, albeit with the lowest number of trainable parameters. In addition, it was shown that the proposed architecture performs a cleaner band-pass, highlighting the necessary frequency bands that were crucial and neurophysiologically plausible in solving the classification tasks.

8

Robotic orthosis compared to virtual hand for Brain–Computer Interface feedback

Cantillo-Negrete Jessica, Carino-Escobar Ruben I., Carrillo-Mora Paul, Barraza-Madrigal José A., Arias-Carrión Oscar

Biocybernetics and Biomedical Engineering

|

2019

|

Vol. 39, no. 2

263--272

EN

Brain–Computer Interfaces (BCI) allow the control of external devices by decoding the users' intentions from their central nervous system. Feedback, one of the main elements of a closed- loop BCI, is used to enhance the user's performance. The present work aimed to compare the effect of two different feedback sources; congruent anatomical visual hand representation and passive hand movement on BCI performance and cortical activations. Electroencephalography of 12 healthy right-handed subjects was recorded to set a BCI activated by right-hand motor imagery. Afterward, the subjects were asked to control the system by imagining the movement. The system provided either visual feedback, shown on a computer screen or kinesthetic feedback, provided by a robotic hand orthosis. Differences in performance and cortical activations were assessed, using classification accuracy and event-related desynchronization/synchronization in μ and β bands, respectively. Performance was significantly better with kinesthetic feedback as it allowed for higher correct classification of motor imagery. Cortical activations in the ipsilateral central channel in μ were different between the two feedback modalities. Our results imply that healthy subjects can achieve a greater degree of control using a motor imagery-based BCI with kinesthetic feedback than with anatomically congruent visual feedback. Furthermore, cortical activation differences show that kinesthetic feedback seems to elicit higher recruitment of sensorimotor cortex brain cells, which probably reflects enhanced local information modulation related to fine motor processing. Therefore, kinesthetic feedback provided by a robotic orthosis could be a more suitable feedback strategy for BCI systems designed for neuromodulation and neurorehabilitation.

9

Classic genetic algorithm vs. genetic algorithm with aggressive mutation for feature selection for a brain-computer interface

Rejer I., Lorenz K.

Przegląd Elektrotechniczny

|

2015

|

R. 91, nr 2

98-102

EN

The classic genetic algorithm has been successfully applied to many optimization problems. However, its usefulness is limited when it comes to feature selection, particularly if a high reduction rate is expected. The algorithm, in its classic version, returns feature sets containing approximately 50% of the total number of features. In order to decrease this rate, a penalty term penalizing individuals of too many features is often added to the fitness function. This solution seems to be reasonable but, as will be shown in this paper, provides only a slight improvement in the reduction rate. In order to obtain a satisfactory classification accuracy and a high reduction rate, not only the fitness function but also other algorithm elements must be reconsidered.

PL

Klasyczny algorytm genetyczny był z powodzeniem stosowany w wielu problemach optymalizacyjnych, jednakże jego użyteczność jest ograniczona w problemach selekcji cech, zwłaszcza jeżeli wymagana jest wysoka stopa redukcji cech. Algorytm, w jego klasycznej wersji, zwraca zbiory cech zawierające około 50% pierwotnej liczby cech. W celu zmniejszenia tej liczby, do funkcji przystosowania algorytmu dołącza się często człon kary, karzący osobniki kodujące zbiory o zbyt dużej liczbie cech. Takie rozwiązanie wydaje się być rozsądne, ale, jak zostanie to przedstawione w artykule pozwala jedynie na niewielką poprawę stopy redukcji. Stąd, w celu uzyskania satysfakcjonującej dokładności klasyfikacji i wysokiej stopy redukcji, nie tylko funkcja przystosowania, ale również inne elementy algorytmu muszą zostać wzięte pod uwagę.