An approach to removing spatially uncorrelated artifacts from EEG recordings

This work deals with elimination artifacts from electroencephalograms (EEGs). Though many methods for solving the problem have been proposed, they either require direct intervention of the researcher, are based on additional measurements (electrooculogram, electrocardiogram, etc.), or cannot remove artifact activity to a sufficient extent. We proposed an automated method that uses of the fact that the electromyogram (EMG) is not correlated and the artifacts of electrode movement at adjacent derivations and is based on the representation of electrical activity in one derivation through activity in other derivations. The analysis of the proposed method using model signals and examples of real EEG recordings has demonstrated its practical effectiveness.     

A two-dimensional approach for lossless EEG compression

In this paper, we study various lossless compression techniques for electroencephalograph (EEG) signals. We discuss a computationally simple pre-processing technique, where EEG signal is arranged in the form of a matrix (2-D) before compression. We discuss a two-stage coder to compress the EEG matrix, with a lossy coding layer (SPIHT) and residual coding layer (arithmetic coding). This coder is optimally tuned to utilize the source memory and the i.i.d. nature of the residual. We also investigate and compare EEG compression with other schemes such as JPEG2000 image compression standard, predictive coding based shorten, and simple entropy coding. The compression algorithms are tested with University of Bonn database and Physiobank Motor/Mental Imagery database. 2-D based compression schemes yielded higher lossless compression compared to the standard vector-based compression, predictive and entropy coding schemes. The use of pre-processing technique resulted in 6% improvement, and the two-stage coder yielded a further improvement of 3% in compression performance.     

A data mining approach for analyzing density maps representing macromolecular structures

Results of electron microscopy-based three-dimensional reconstructions of macromolecules or their complexes are usually stored as density maps. Each point ("voxel") in the map represents a density value and one approach for studying details of the map is to display an isosurface enclosing areas of interest. We have taken a data mining approach not only focusing on the areas of immediate interest but determining all possible separate entities ("blobs") from a density map. After the entire density map is analyzed with our mining program BLOBBER, properties of all detected blobs can be browsed and sets of blobs can be visualized using our VIZBLOB program. Since BLOBBER analyzes density maps using only density information and relates it to spatial relationships, BLOBBER can be used to analyze symmetrical or asymmetrical density maps from any source. To test our program we have analyzed published bacteriophage PRD1 reconstructions. We identified various structural details ranging from individual proteins to major complexes such as the whole capsid shell and more elaborate details of possible connections between membrane interfaces. This approach can also be a useful preprocessing tool for visualizing reconstructions.     

An analysis of spontaneous human cortical EEG activity to odours

Using single trial presentation of odours, mapping of spontaneousEEG was made in naive subjects. Systematic increases in EEGamplitudes for alpha (8–13 Hz) frequency were shown forthe odours used and discrimination of odours was shown by specificelectrodes. These were mainly in the area bounded by International10/20 scalp locations CP1, PZ, CP2, TCP2 and C4. It was alsoshown that EEG recordings from more anterior electrodes couldbe related to psychometric responses. The findings showed thatstatistically reliable electrical amplitudes to odour presentationswere discernible in the noisy EEG environment on the surfaceof the brain.     

A multi-context learning approach for EEG epileptic seizure detection

Background

Epilepsy is a neurological disease characterized by unprovoked seizures in the brain. The recent advances in sensor technologies allow researchers to analyze the collected biological records to improve the treatment of epilepsy. Electroencephalogram (EEG) is the most commonly used biological measurement to effectively capture the abnormalities of different brain areas during the EEG seizures. To avoid manual visual inspection from long-term EEG readings, automatic epileptic EEG seizure detection has become an important research issue in bioinformatics.

Results

We present a multi-context learning approach to automatically detect EEG seizures by incorporating a feature fusion strategy. We generate EEG scalogram sequences from the EEG records by utilizing waveform transform to describe the frequency content over time. We propose a multi-stage unsupervised model that integrates the features extracted from the global handcrafted engineering, channel-wise deep learning, and EEG embeddings, respectively. The learned multi-context features are subsequently merged to train a seizure detector.

Conclusions

To validate the effectiveness of the proposed approach, extensive experiments against several baseline methods are carried out on two benchmark biological datasets. The experimental results demonstrate that the representative context features from multiple perspectives can be learned by the proposed model, and further improve the performance for the task of EEG seizure detection.

     

An alpha-numeric code for representing N-linked glycan structures in secreted glycoproteins

Advances in high-throughput techniques have led to the creation of increasing amounts of glycome data. The storage and analysis of this data would benefit greatly from a compact notation for describing glycan structures that can be easily stored and interpreted by computers. Towards this end, we propose a fixed-length alpha-numeric code for representing N-linked glycan structures commonly found in secreted glycoproteins from mammalian cell cultures. This code, GlycoDigit, employs a pre-assigned alpha-numeric index to represent the monosaccharides attached in different branches to the core glycan structure. The present branch-centric representation allows us to visualize the structure while the numerical nature of the code makes it machine readable. In addition, a difference operator can be defined to quantitatively differentiate between glycan structures for further analysis. The usefulness and applicability of GlycoDigit were demonstrated by constructing and visualizing an N-linked glycosylation network.     

EEG reactions during creative activity

Evoked activity in response to light was recorded in students performing a verbal creative task. The changes in the amplitude of the N200 negative component of evoked potentials were analyzed. The amplitude of the N200 component was significantly increased in the frontal and anterior frontal areas of the left hemisphere, which indicated increased activity in the cortical structures involved in divergent thinking. The amplitude of the N200 component was increased in the temporo-parieto-occipital area of the right hemisphere, which indicated that the posterior associative region of the right hemisphere was also involved in the creative activity. The data obtained suggest that the frontal and temporo-parieto-occipital areas of the cerebral cortex actively participate in the performance of a creative test with distinct elements of complexity.     

An interval approach for dealing with flux distributions and elementary modes activity patterns

This work introduces the use of an interval representation of fluxes. This representation can be useful in two common situations: (a) when fluxes are uncertain due to the lack of accurate measurements and (b) when the flux distribution is partially unknown. In addition, the interval representation can be used for other purposes such as dealing with inconsistency or representing a range of behaviour. Two main problems are addressed. On the one hand, the translation of a metabolic flux distribution into an elementary modes or extreme pathways activity pattern is analysed. In general, there is not a unique solution for this problem but a range of solutions. To represent the whole solution region in an easy way, it is possible to compute the alpha-spectrum (i.e., the range of possible values for each elementary mode or extreme pathway activity). Herein, a method is proposed which, based on the interval representation of fluxes, makes it possible to compute the alpha-spectrum from an uncertain or even partially unknown flux distribution. On the other hand, the concept of the flux-spectrum is introduced as a variant of the metabolic flux analysis methodology that presents some advantages: applicable when measurements are insufficient (underdetermined case), integration of uncertain measurements, inclusion of irreversibility constraints and an alternative procedure to deal with inconsistency. Frequently, when applying metabolic flux analysis the available measurements are insufficient and/or uncertain and the complete flux distribution cannot be uniquely calculated. The method proposed here allows the determination of the ranges of possible values for each non-calculable flux, resulting in a flux region called flux-spectrum. In order to illustrate the proposed methods, the example of the metabolic network of CHO cells cultivated in stirred flasks is used.     

Analysis of phase-locking is informative for studying event-related EEG activity

 A new method is presented for quantitative evaluation of single-sweep phase and amplitude electroencephalogram (EEG) characteristics that is a more informative approach in comparison with conventional signal averaging. In the averaged potential, phase-locking and amplitude effects of the EEG response cannot be separated. To overcome this problem, single-trial EEG sweeps are decomposed into separate presentations of their phase relationships and amplitude characteristics. The stability of the phase-coupling to stimulus is then evaluated independently by analyzing the single-sweep phase presentations. The method has the following advantages: information about stability of the phase-locking can be used to assess event-related oscillatory activity; the method permits evaluation of the timing of event-related phase-locking; and a global assessment and comparison of the phase-locking of ensembles of single sweeps elicited in different processing conditions is possible. The method was employed to study auditory alpha and theta responses in young and middle-aged adults. The results showed that whereas amplitudes of frequency responses tended to decrease, the phase-locking increased significantly with age. The synchronization with stimulus (phase-locking) was the only parameter reliably to differentiate the brain responses of the two age groups, as well as to reveal specific age-related changes in frontal evoked alpha activity. Thus, the present approach can be used to evaluate dynamic brain processes more precisely. Received: 12 February 1996 / Accepted in revised form: 11 October 1996     

Graphical methods for representing computer classifications

An account is given of a method of representing three-dimensional configurations in two dimensions by the use of sections and contours, and of its application to numerical classification.     

Predictability of human EEG: a dynamical approach

The electroencephalogram recordings from human scalp are analysed in the framework of recent methods of nonlinear dynamics. Three stages of brain activity are considered: the alpha waves (eyes closed), the deep sleep (stage four) and the Creutzfeld-Jakob coma. Two dynamical parameters of the attractors are evaluated. These are the Lyapunov exponents, which measure the divergence or convergence of trajectories in phase space and the Kolmogorov or metric entropy, whose inverse gives the mean predicting time of a given EEG signal. In all the stages considered, the results reveal the presence of at least two positive Lyapunov exponents, which are the footprints of chaos. This number increases to three positive exponents in the case of alpha waves, indicating that although for very short episodes the alpha waves seem extremely coherent, the variability of the brain increases markedly over larger periods of activity. The degree of entropy/chaos increases from coma to deep sleep and then to alpha waves. The large predicting time observed for deep sleep suggests that these waves are related to a slow rate of information processing. The predicting time of the alpha waves is much smaller, indicating a rapid loss of information. Finally, with the help of the Lyapunov exponents, the attractor's dimensions are evaluated using two different conjectures and compared to values obtained previously by the Grassberger-Procaccia algorithm.     

An empirical EEG analysis in brain death diagnosis for adults

Electroencephalogram (EEG) is often used in the confirmatory test for brain death diagnosis in clinical practice. Because EEG recording and monitoring is relatively safe for the patients in deep coma, it is believed to be valuable for either reducing the risk of brain death diagnosis (while comparing other tests such as the apnea) or preventing mistaken diagnosis. The objective of this paper is to study several statistical methods for quantitative EEG analysis in order to help bedside or ambulatory monitoring or diagnosis. We apply signal processing and quantitative statistical analysis for the EEG recordings of 32 adult patients. For EEG signal processing, independent component analysis (ICA) was applied to separate the independent source components, followed by Fourier and time-frequency analysis. For quantitative EEG analysis, we apply several statistical complexity measures to the EEG signals and evaluate the differences between two groups of patients: the subjects in deep coma, and the subjects who were categorized as brain death. We report statistically significant differences of quantitative statistics with real-life EEG recordings in such a clinical study, and we also present interpretation and discussions on the preliminary experimental results.

Time representing cortical activities: two models inspired by prefrontal persistent activity

 Timing information in the range of seconds is significantly correlated with our behavior. There is growing interest in the cognitive behaviors that rely on perception, comparison, or generation of timing. However, little is known about the neural mechanisms underlying such behaviors. Here we model two different neural mechanisms to represent timing information in the range of seconds. In one model, a recurrent network of bistable spiking neurons shows a quasistable state that is initiated by a brief input and typically lasts for a few to several seconds. The duration of this quasistable activity may be regarded as the neural representation of internal time obeying a psychophysical law of time recognition. Another model uses synfire chains to provide the timing information necessary for predicting the times of anticipated events. In this model, the neurons projected to by multiple synfire chains are conditioned to fire synchronously at the times when an external event (GO signal) is expected. The conditioning is accomplished by spike-timing-dependent plasticity. The two models are inspired by the prefrontal activities of the monkeys engaging in different timing-information-related tasks. Thus, this cortical region may provide the timing information required for organizing various behaviors. Received: 12 March 2002 / Accepted in revised form: 26 November 2002 / Published online: 28 March 2003 Correspondence to: T. Fukai (e-mail: tfukai@eng.tamagawa.ac.jp, Tel.: +81-42-7398434, Fax: +81-42-7397135) Acknowledgements. K. Kitano was supported by Japan Society for the Promotion of Science.     

Cytogenetic activity of metabolites of pesticides representing different classes of chemical compounds

The cytogenetic activity of some substances formed in agricultural plants during metabolism of pesticides of four classes of chemical compounds was studied in the culture of human peripheric blood lymphocytes. Metabolites were shown either to have mutagenic properties similar to those of the initial compounds (ziramtetramethylthiourea, both being mutagens; captan-phthalimide, both possessing no cytogenetic activity) or to be considerably transformed in comparison with them as a result of deactivation (benomile-MBC) or activation (betanal-MHPC) processes. The latter variant if being determined for the genetic hazard of the pesticide necessitates to take into account data on the mutagenic character of those metabolites which really might enter the human organism.