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
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
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.
3
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
To investigate the optimal filter settings for pre-processing of Movement Related Cortical Potentials (MRCP) for the detection through EEG in single trial, we have proposed a novel Non-Linear Optimized Spatial Filter (NL-SF) and compared it to the Optimized Spatial Filtering (OSF) used in literature. MRCPs from EEG recordings are emphasized, calculating the optimal non-linear combination of channels which isolates the signal of interest. The method is applied to EEG data recorded from 16 healthy patients either executing or imagining 50 self-paced upper limb movements (palmar grasp). MRCPs have been identified from the outputs of the two filters by matching with a template built by averaging responses to movement intentions in the training set. NL-SF had a median accuracy on the overall dataset of 84.6%, which is significantly better than that of OSF (i.e., 76.9%). Being a filter and feasible for self-paced applications, it could be of interest in online BCI system design.
4
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
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
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Electroencephalography (EEG) is the signal of intrigue that has immense application in the clinical diagnosis of various neurological, psychiatric, psychological, psychophysiological, and neurocognitive disorders. It is significantly crucial in neural communication, brain-computer interface, and other practical tasks. EEG signal is exceptionally susceptible to artifacts, which are external noise signals originated from non-cerebral regions. The interference of artifacts in EEG signals can potentially affect the original recorded EEG signal quality and pattern. Therefore, artifact removal from EEG signal is critically important before applying it to a specific task for accurate outcomes. Researchers have proposed numerous techniques to remove various artifacts present in the contaminated EEG signal. However, neither optimum method nor criterion stands standard for endorsement of clinically recorded EEG signals. Therefore, the research related to artifact elimination from EEG signal is challenging and perplexing task. This paper attempts to give an extensive outline of the advancement in methodologies to eliminate one of the most common artifacts, i.e., ocular artifact. It is anticipated that the study will enlighten the researchers on all the existing ocular artifact elimination techniques with a validated simulation model on the recorded EEG signal. In future advancements, Standard norms in artifact elimination techniques are expected to diminish the neurologist’s load by substantiating the clinical diagnosis after gaining correct information from artifact-free EEG signals.
A Brain‐Computer Interface (BCI) is an instrument capa‐ ble of commanding machine with brain signal. The mul‐ tiple types of signals allow designing many applications like the Oddball Paradigms with P300 signal. We propose an EEG classification system applied to BCI using the con‐ volutional neural network (ConvNet) for P300 problem. The system consists of three stages. The first stage is a Spatiotemporal convolutional layer which is a succession of temporal and spatial convolutions. The second stage contains 5 standard convolutional layers. Finally, a lo‐ gistic regression is applied to classify the input EEG sig‐ nal. The model includes Batch Normalization, Dropout, and Pooling. Also, It uses Exponential Linear Unit (ELU) function and L1‐L2 regularization to improve the lear‐ ning. For experiments, we use the database Dataset II of the BCI Competition III. As a result, we get an F1‐score of 53.26% which is higher than the BN3 model.
7
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Steady-state visual evoked potential (SSVEP) based brain–computer interfaces have been widely studied because these systems have potential to restore capabilities of communication and control of disable people. Identifying target frequency using SSVEP signals is still a great challenge due to the poor signal-to-noise ratio of these signals. Commonly, this task is carried out with detection algorithms such as bank of frequency-selective filters and canonical correlation analysis. This work proposes a novel method for the detection of SSVEP that combines the empirical mode decomposition (EMD) and a power spectral peak analysis (PSPA). The proposed EMD+PSPA method was evaluated with two EEG datasets, and was compared with the widely used FB and CCA. The first dataset is freely available and consists of three flickering light sources; the second dataset was constructed and consists of six flickering light sources. The results showed that proposed method was able to detect SSVEP with high accuracy (93.67 ± 9.97 and 78.19 ± 23.20 for the two datasets). Furthermore, the detection accuracy results achieved with the first dataset showed that EMD+PSPA provided the highest detection accuracy (DA) in the largest number of participants (three out of five), and that the average DA across all participant was 93.67 ± 9.97 which is 7% and 4% more than the average DA achieved with FB and CCA, respectively.
8
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Detection of eye closing/opening from alpha-blocking in the EEG of occipital region has been used to build human-machine interfaces. This paper presents an alternative method for detection of eye closing/opening from EOG signals in an online setting. The accuracies for correct detection of eye closing and opening operations with the proposed techniques were found to be 95.6% and 91.9% respectively for 8 healthy subjects. These techniques were then combined with the detection of eye blinks, the accuracy of which turned out to be 96.9%. This was then used to build an interface for robotic arm control for a pick and place task. The same task was also carried out using a haptic device as a master. The speed and accuracy for these two methods were then compared to assess quantitatively the ease of using this interface. It appears that the proposed interface will be very useful for persons with neurodegenerative disorders who can perform eye closing/opening and eye blinks.
9
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
In this study, a brain–computer interface (BCI) system known as P300 speller is used to spell the word or character without any muscle activity. For P300 signal classification, feature extraction is an important step. In this work, deep feature learning techniques based on sparse autoencoder (SAE) and stacked sparse autoencoder (SSAE) are proposed for feature extraction. Deep feature provides the abstract information about the signal. This work proposes fusion of deep features with the temporal features, which provides abstract and temporal information about the EEG signal. These deep feature and temporal feature are partially complement of each other to represent the EEG signal. For classification of the EEG signal, an ensemble of support vector machines (ESVM) is adopted as it helps to reduce the classifiers variability. In classifier ensemble system, the score of individual classifier is not at the same level. To transform these scores into a common level, min–max normalization is proposed prior to combining them. Min-max normalization scales the classifiers' score between 0 and 1. The experiments are conducted on three standard public datasets, dataset IIb of BCI Competition II, dataset II of the BCI Competition III and BNCI Horizon dataset. The experimental results show that the proposed method yields better or comparable performance compared to earlier reported techniques.
10
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Brain–computer interfaces based on steady-state visual evoked potentials have recently gained increasing attention due to high performance and minimal user training. Stimulus frequencies in the range of 4–60 Hz have been used in these systems. However, eye fatigue when looking at low-frequency flickering lights, higher risk of induced epileptic seizure for medium-frequency flickers, and low signal amplitude for high-frequency flickers complicate appropriate selection of flickering frequencies. Here, different flicker frequencies were evaluated for development of a brain–computer interface speller that ensures user's comfort as well as the system's efficiency. A frequency detection algorithm was also proposed based on Least Absolute Shrinkage and Selection Operator estimate that provides excellent accuracy using only a single channel of EEG. After evaluation of the SSVEP responses in the range of 6–60 Hz, three stimulus frequency sets of 30–35, 35–40 and 40–45 Hz were adopted and the system's performance and corresponding eye fatigue were compared. While the accuracy of the asynchronous speller for all three stimulus frequency sets was close to the maximum (average 97.6%), repeated measures ANOVA demonstrated that the typing speed for 30–35 Hz (8.09 char/min) and 35–40 Hz (8.33 char/min) are not significantly different, but are significantly higher than for 40–45 Hz (6.28 char/min). On the other hand, the average eye fatigue scale for 35–40 Hz (80%) is comparable to that for 40–45 Hz (85%), but very higher than for 30–35 Hz (60%). Therefore, 35–40 Hz range was proposed for the system which resulted in 99.2% accuracy and 67.1 bit/min information transfer rate.
11
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Brain Computer Interface (BCI) generate a direct method to communicate with the outside world. Many patients are not able to communicate. For example:- the patient who are suffered with the several disease like post stroke - the process of thinking, remembering \& recognizing can be challenging. Because of spinal cord injuries or brain stem stroke the patient loss the monitoring power. EEG based brain computer interface (BCI) feature is beneficial to scale the brain movement \& convert them into a instruction for monitoring. In this paper our objective is to study about various applications of EEG based signal of the different disease like spinal cord injury, post stroke and ALS (amyotrophic lateral sclerosis) etc.
12
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
P300 speller-based brain-computer interface (BCI) allows a person to communicate with a computer using only brain signals. In order to achieve better reliability and user continence, it is desirable to have a system capable of providing accurate classification with as few EEG channels as possible. This article proposes an approach based on multi-objective binary differential evolution (MOBDE) algorithm to optimize the system accuracy and number of EEG channels used for classification. The algorithm on convergence provides a set of pareto-optimal solutions by solving the trade-off between the classification accuracy and the number of channels for Devanagari script (DS)-based P300 speller system. The proposed method is evaluated on EEG data acquired from 9 subjects using a 64 channel EEG acquisition device. The statistical analysis carried out in the article, suggests that the proposed method not only increases the classification accuracy but also increases the over-all system reliabil-ity in terms of improved user-convenience and information transfer rate (ITR) by reducing the EEG channels. It was also revealed that the proposed system with only 16 channels was able to achieve higher classification accuracy than a system which uses all 64 channel's data for feature extraction and classification.
13
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
To construct brain–computer interface (BCI), an event-related potential (ERP) induced by a tactile stimulus is investigated in this paper. For ERP-based BCI, visual or auditory information is frequently used as the stimulus. In the present study, we focus on tactile sensations to reserve their visual and auditory senses for other activities. Several patterns of mechanical tactile stimulation were applied to the index fingers of both hands using two piezo actuators that were used as a braille display. Human experiments based on the oddball paradigm were carried out. Subjects were instructed to pay attention to unusual target stimuli while avoiding other frequent non-target stimuli. The extracted features were classified by applying stepwise linear discriminant analysis. As a result, an accuracy of 85% and 60% were obtained for 2- and 4-class classification, respectively. The accuracy was improved by increasing the number of electrodes even when short stimulus intervals were used.
14
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
A very interesting research goal is to find underlying sources generating the EEG signal–referred to as the ‘‘EEG inverse problem’’. Its aim is to determine spatial distribution of brain activity, described by local brain currents density, on the basis of potentials measured on the scalp as EEG signal. The purpose of the research presented in the article was to check whether the results of the inverse problem solution, obtained by the LORETA algorithm for the reduced set of 8 electrodes selected by the authors will be close to the results for the initial set of 32 electrodes. EEG signals were registered during the BCI operation based on ERD/ERS potentials. Obtained results showed no significant differences in the location of the most important sources in both cases. It is worth emphasizing that reducing the number of electrodes would have a significant impact on an BCI ergonomics.
16
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Wirelessly networked systems of sensors could enable revolutionary applications at the intersection of biomedical science, networking and control systems. It has a strong potential to take ahead the applications of wireless sensor networks. In this paper, a wireless brain computer interface (BCI) framework for drowsiness detection is proposed, which uses electroencephalogram (EEG) signals produced from the brain wave sensors. The proposed BCI framework comprises of a braincap containing EEG sensors, wireless signal acquisition unit and a signal processing unit. The signal processing unit continuously monitor the preprocessed EEG signals and to trigger a warning tone if a drowsy state happens. This experimental setup provides longer time EEG monitoring and drowsiness detection by incorporating the clustering mechanism into the wireless networks.
The article presents the results of research in controlling the mobile application with the EEG signals and eye blinking. Authors proposed a prototype solution of a brain-computer interface that can be used by people with total motor impairment to control chosen mobile application on their mobile phone. There was a NeuroSky MindWave Mobile device used during experiments. Two software tools for mobile devices were specially implemented. First one helps to analyse the EEG signals and recognize eye blinks, second one - interprets them and executes assigned actions. Different configurations of settings were used during the studies. They included: single blink or double blink, level of focus, period of focus. Experiments results show that a man equipped with a personal EEG sensor and eye blinking detector can remotely touchless use mobile applications installed on smartphones or tablets.
Brain computer interface (BCI) is a system allows a user to control external devices or to communicate with other people using only his or her thoughts. The P300 speller is one such BCI in which users input letters. For inputting letters via the P300 speller, higher accuracy and shorter input times are needed, especially given densely populated display screens. We propose a new interface with a second display in the P300 speller that the user can switch to and from by selecting the “next” or “back” commands, therby reducing the density of displayed letters and improving the performance of the P300 speller. We show the comparison results in terms of accuracy and input times between the conventional interface and proposed interface.
One source of EEG data quality deterioration is noise. The others are artifacts, such as the eye blinking, oculogyration, heart beat, or muscle activity. All these factors mentioned above contribute to the disappointing and poor quality of EEG signals. There are some solutions which allow increase of this signals quality. One of them is Common Spatial Patterns. Some scientific papers report that CSP can only be effectively used if there are many electrodes available. The aim of this paper is to use CSP method applied in the process of creating a brain computer interface in order to find out if there are any benefits of using this method in 3 channels BCI system.
There are a lot of problems that arise in the process of building a brain-computer interface based on electroencephalographic signals (EEG). A huge imbalance between a number of experiments possible to conduct and the size of feature space, containing features extracted from recorded signals, is one of them. To reduce this imbalance, it is necessary to apply methods for feature selection. One of the approaches for feature selection, often taken in brain-computer interface researches, is a classic genetic algorithm that codes all features within each individual. In this study, there will be shown, that although this approach allows obtaining a set of features of high classification precision, it also leads to a feature set highly redundant comparing to a set of features selected using a forward selection method or a genetic algorithm equipped with individuals of a given (very small) number of genes.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.