scispace - formally typeset
Journal ArticleDOI: 10.1109/LSP.2020.3006417

Wavelet Based Waveform Distortion Measures for Assessment of Denoised EEG Quality With Reference to Noise-Free EEG Signal

03 Jul 2020-IEEE Signal Processing Letters (IEEE)-Vol. 27, pp 1260-1264
Abstract: An objective distortion measure is very crucial to accurately quantify the distortion introduced in the electroencephalogram (EEG) signal during the denoising process Most of the existing algorithms report their denoising performance by comparing the original EEG signal and the reconstructed EEG signal using root mean square error (RMSE) and other similar measures However, it is very important to quantify the distortion in each band of EEG signal since each band provides distinct information about the specific brain activity Furthermore, quantification of band-wise distortion enables the selection of particular denoising algorithm for the application-specific EEG signal analysis Therefore, in this paper, we propose two robust distortion measures such as weighted signal to noise ratio (WSNR) and weighted correlation coefficient (WCC) for accurately representing the objective reconstruction loss in each band These performance measures are computed between the wavelet subbands of the original and the denoised/reconstructed signal with weights equal to the relative wavelet energy and wavelet entropy of the corresponding subband wavelet coefficients To demonstrate the effectiveness of the proposed performance measures, we evaluate the performance of six existing denoising methods using these measures Results depict that these measures can adequately provide high mutual agreement between objective scores and subjective analysis more

Topics: Wavelet (59%), Distortion (58%), Noise (signal processing) (54%) more

Open accessPosted Content
08 Feb 2021-arXiv: Learning
Abstract: The cross-subject application of EEG-based brain-computer interface (BCI) has always been limited by large individual difference and complex characteristics that are difficult to perceive Therefore, it takes a long time to collect the training data of each user for calibration Even transfer learning method pre-training with amounts of subject-independent data cannot decode different EEG signal categories without enough subject-specific data Hence, we proposed a cross-subject EEG classification framework with a generative adversarial networks (GANs) based method named common spatial GAN (CS-GAN), which used adversarial training between a generator and a discriminator to obtain high-quality data for augmentation A particular module in the discriminator was employed to maintain the spatial features of the EEG signals and increase the difference between different categories, with two losses for further enhancement Through adaptive training with sufficient augmentation data, our cross-subject classification accuracy yielded a significant improvement of 1585% than leave-one subject-out (LOO) test and 857% than just adapting 100 original samples on the dataset 2a of BCI competition IV Moreover, We designed a convolutional neural networks (CNNs) based classification method as a benchmark with a similar spatial enhancement idea, which achieved remarkable results to classify motor imagery EEG data In summary, our framework provides a promising way to deal with the cross-subject problem and promote the practical application of BCI more

1 Citations

Journal ArticleDOI: 10.1109/TIM.2021.3071217
Aiping Liu1, Gongzheng Song1, Soojin Lee2, Xueyang Fu1  +1 moreInstitutions (2)
Abstract: Electroencephalography (EEG) is an important noninvasive neural recording technique with a broad application in the field of neurological instrumentation and measurement. However, EEG signals are often contaminated by muscle artifacts, adversely affecting the subsequent analysis. Joint blind source separation (JBSS) models have been successfully applied to remove muscle artifacts from EEG recordings, although most of them were designed for EEG collected in well-controlled conditions. Without considering the dynamics of underlying mixtures in complex environments may hinder the real mobile and long-term healthcare monitoring. To deal with such concern, we assume that the mixing process of sources dynamically changes over time and propose a state-dependent JBSS model by integrating the hidden Markov model with independent vector analysis in a maximum likelihood framework. It is capable of identifying the varying sources of muscle artifact components and underlying EEG signals. The proposed method was evaluated on both simulated and semi-simulated data, and demonstrated superior performance compared with other popular approaches for muscle artifacts removal in dynamic environments. The state-dependent JBSS model provides a novel way to investigate the temporal dynamics of multiple multidimensional biomedical data sets simultaneously. more

Proceedings ArticleDOI: 10.1109/ICOICT52021.2021.9527426
03 Aug 2021-
Abstract: Electroencephalography (EEG) is a technique for measuring electrical activity on the scalp. The EEG detects voltage fluctuations caused by ion currents in brain neurons. The brain-computer interface system (BCIs) is intended to enable humans to monitor machines and interact with computers through their brains. It intends to construct non-muscular production pathways to convert brain function into discriminatory control commands correlated with various EEG signals dependent on motorized image patterns. Research on EEG is currently growing, especially in the field of motor imaging. EEG signal processing would be a feasible option for developing such a BCI device. The four basic stages of classical BCI are multi-class EEG signal acquisition, signal preprocessing, feature extraction, and motor imagery classification based on EEG. This study aims to determine the effect of wavelet packet decomposition (WPD) and common spatial pattern (CSP) feature extraction to optimize feature selection using the ensemble learning method. The method used in this research is experimental, where the stages begin with preprocessing, feature extraction with WPD and CSP, classification using ensemble learning and implementing feature selection using the principal component analysis (PCA) and select from the model (SFM). The results are the comparison of the accuracy generated from each method, including random forest (RF) of 74.71%, random forest with principal component analysis (RFPCA) of 68.01%, random forest with select from the model (RFSFM) of 82.15%, extra trees (ET) of 77.97%, extra trees with principal component analysis (ETPCA) of 64.18% and extra trees with selected from the model (ETSFM) of 83.28%. more

Topics: Feature extraction (56%), Ensemble learning (55%), Random forest (54%) more

Proceedings ArticleDOI: 10.1109/NCC52529.2021.9530053
Payal Patel1, Udit Satija2Institutions (2)
27 Jul 2021-
Abstract: Recently, convolutional neural network (CNN) has played a crucial role in classifying epileptic seizures due to its capability of automatically learning the discriminatory features from the raw electroencephalogram (EEG) data. Moreover, most of the existing methods considered artifact-free EEG data for extracting features. In this paper, we analyze the impact of ocular artifacts on the performance of CNN in extracting reliable features from the EEG data for seizure classification. Furthermore, we also analyze the robustness of CNN in determining the accurate and reliable features not only from raw EEG data but also from spectral domain EEG data. The performance of the method is evaluated on the EEG signals taken from the Bonn's dataset with different types and levels of ocular artifacts. Performance evaluation results demonstrate that the classification accuracy of the method is degraded significantly under the presence of ocular artifacts. Furthermore, it is observed that the proposed CNN architecture is able to extract the discriminatory features from spectral EEG data more accurately as compared to the raw temporal EEG data. more


Open accessJournal ArticleDOI: 10.1109/34.192463
Stéphane Mallat1Institutions (1)
Abstract: Multiresolution representations are effective for analyzing the information content of images. The properties of the operator which approximates a signal at a given resolution were studied. It is shown that the difference of information between the approximation of a signal at the resolutions 2/sup j+1/ and 2/sup j/ (where j is an integer) can be extracted by decomposing this signal on a wavelet orthonormal basis of L/sup 2/(R/sup n/), the vector space of measurable, square-integrable n-dimensional functions. In L/sup 2/(R), a wavelet orthonormal basis is a family of functions which is built by dilating and translating a unique function psi (x). This decomposition defines an orthogonal multiresolution representation called a wavelet representation. It is computed with a pyramidal algorithm based on convolutions with quadrature mirror filters. Wavelet representation lies between the spatial and Fourier domains. For images, the wavelet representation differentiates several spatial orientations. The application of this representation to data compression in image coding, texture discrimination and fractal analysis is discussed. > more

Topics: Wavelet (65%), Wavelet transform (65%), Orthogonal wavelet (65%) more

19,033 Citations

Open accessJournal ArticleDOI: 10.1161/01.CIR.101.23.E215
13 Jun 2000-Circulation
Abstract: —The newly inaugurated Research Resource for Complex Physiologic Signals, which was created under the auspices of the National Center for Research Resources of the National Institutes of He... more

8,656 Citations

Journal ArticleDOI: 10.1111/1469-8986.3720163
Tzyy-Ping Jung1, Tzyy-Ping Jung2, Scott Makeig2, Colin Humphries1  +6 moreInstitutions (2)
01 Mar 2000-Psychophysiology
Abstract: Eye movements, eye blinks, cardiac signals, muscle noise, and line noise present serious problems for electroencephalographic (EEG) interpretation and analysis when rejecting contaminated EEG segments results in an unacceptable data loss. Many methods have been proposed to remove artifacts from EEG recordings, especially those arising from eye movements and blinks. Often regression in the time or frequency domain is performed on parallel EEG and electrooculographic (EOG) recordings to derive parameters characterizing the appearance and spread of EOG artifacts in the EEG channels. Because EEG and ocular activity mix bidirectionally, regressing out eye artifacts inevitably involves subtracting relevant EEG signals from each record as well. Regression methods become even more problematic when a good regressing channel is not available for each artifact source, as in the case of muscle artifacts. Use of principal component analysis (PCA) has been proposed to remove eye artifacts from multichannel EEG. However, PCA cannot completely separate eye artifacts from brain signals, especially when they have comparable amplitudes. Here, we propose a new and generally applicable method for removing a wide variety of artifacts from EEG records based on blind source separation by independent component analysis (ICA). Our results on EEG data collected from normal and autistic subjects show that ICA can effectively detect, separate, and remove contamination from a wide variety of artifactual sources in EEG records with results comparing favorably with those obtained using regression and PCA methods. ICA can also be used to analyze blink-related brain activity. more

2,681 Citations

Journal ArticleDOI: 10.1103/PHYSREVE.64.061907
20 Nov 2001-Physical Review E
Abstract: We compare dynamical properties of brain electrical activity from different recording regions and from different physiological and pathological brain states Using the nonlinear prediction error and an estimate of an effective correlation dimension in combination with the method of iterative amplitude adjusted surrogate data, we analyze sets of electroencephalographic (EEG) time series: surface EEG recordings from healthy volunteers with eyes closed and eyes open, and intracranial EEG recordings from epilepsy patients during the seizure free interval from within and from outside the seizure generating area as well as intracranial EEG recordings of epileptic seizures As a preanalysis step an inclusion criterion of weak stationarity was applied Surface EEG recordings with eyes open were compatible with the surrogates' null hypothesis of a Gaussian linear stochastic process Strongest indications of nonlinear deterministic dynamics were found for seizure activity Results of the other sets were found to be inbetween these two extremes more

Topics: Electroencephalography (54%), Surrogate data (52%)

1,973 Citations

Journal ArticleDOI: 10.1016/S0165-0270(00)00356-3
Osvaldo A. Rosso1, S. Blanco1, Juliana Yordanova2, Vasil Kolev2  +3 moreInstitutions (3)
Abstract: Since traditional electrical brain signal analysis is mostly qualitative, the development of new quantitative methods is crucial for restricting the subjectivity in the study of brain signals. These methods are particularly fruitful when they are strongly correlated with intuitive physical concepts that allow a better understanding of brain dynamics. Here, new method based on orthogonal discrete wavelet transform (ODWT) is applied. It takes as a basic element the ODWT of the EEG signal, and defines the relative wavelet energy, the wavelet entropy (WE) and the relative wavelet entropy (RWE). The relative wavelet energy provides information about the relative energy associated with different frequency bands present in the EEG and their corresponding degree of importance. The WE carries information about the degree of order/disorder associated with a multi-frequency signal response, and the RWE measures the degree of similarity between different segments of the signal. In addition, the time evolution of the WE is calculated to give information about the dynamics in the EEG records. Within this framework, the major objective of the present work was to characterize in a quantitative way functional dynamics of order/disorder microstates in short duration EEG signals. For that aim, spontaneous EEG signals under different physiological conditions were analyzed. Further, specific quantifiers were derived to characterize how stimulus affects electrical events in terms of frequency synchronization (tuning) in the event related potentials. more

Topics: Wavelet (61%), Discrete wavelet transform (58%), Entropy (energy dispersal) (55%) more

709 Citations

No. of citations received by the Paper in previous years
Network Information
Related Papers (5)
26 Jul 2017

Zaid Abdi Alkareem Alyasseri, Ahamad Tajudin Khader +1 more

11 Sep 2020

Liwei Cheng, Duanling Li +3 more

26 Jul 2017

Zaid Abdi Alkareem Alyasseri, Ahamad Tajudin Khader +1 more

01 Jan 2021

Zaid Abdi Alkareem Alyasseri, Ahamad Tajudin Khader +3 more