Correlation dimension based lossless compression of EEG signals

Transmission of long duration EEG signals without loss of information is essential for telemedicine based applications. In this work, a lossless compression scheme for EEG signals based on neural network predictors using the concept of correlation dimension (CD) is proposed. EEG signals which are considered as irregular time series of chaotic processes can be characterized by the non-linear dynamic parameter CD which is a measure of the correlation among the EEG samples. The EEG samples are first divided into segments of 1 s duration and for each segment, the value of CD is calculated. Blocks of EEG samples are then constructed such that each block contains segments with closer CD values. By arranging the EEG samples in this fashion, the accuracy of the predictor is improved as it makes use of highly correlated samples. As a result, the magnitude of the prediction error decreases leading to less number of bits for transmission. Experiments are conducted using EEG signals recorded under different physiological conditions. Different neural network predictors as well as classical predictors are considered. Experimental results show that the proposed CD based preprocessing scheme improves the compression performance of the predictors significantly.     

Relevant EEG features for the classification of spontaneous motor-related tasks

 There is a growing interest in the use of physiological signals for communication and operation of devices for the severely motor disabled as well as for healthy people. A few groups around the world have developed brain-computer interfaces (BCIs) that rely upon the recognition of motor-related tasks (i.e., imagination of movements) from on-line EEG signals. In this paper we seek to find and analyze the set of relevant EEG features that best differentiate spontaneous motor-related mental tasks from each other. This study empirically demonstrates the benefits of heuristic feature selection methods for EEG-based classification of mental tasks. In particular, it is shown that the classifier performance improves for all the considered subjects with only a small proportion of features. Thus, the use of just those relevant features increases the efficiency of the brain interfaces and, most importantly, enables a greater level of adaptation of the personal BCI to the individual user. Received: 15 January 2001 / Accepted in revised form: 19 July 2001     

Relevance of nonlinear lumped-parameter models in the analysis of depth-EEG epileptic signals

In the field of epilepsy, the analysis of stereoelectroencephalographic (SEEG, intra-cerebral recording) signals with signal processing methods can help to better identify the epileptogenic zone, the area of the brain responsible for triggering seizures, and to better understand its organization. In order to evaluate these methods and to physiologically interpret the results they provide, we developed a model able to produce EEG signals from “organized” networks of neural populations. Starting from a neurophysiologically relevant model initially proposed by Lopes Da Silva et al. [Lopes da Silva FH, Hoek A, Smith H, Zetterberg LH (1974) Kybernetic 15: 27–37] and recently re-designed by Jansen et al. [Jansen BH, Zouridakis G, Brandt ME (1993) Biol Cybern 68: 275–283] the present study demonstrates that this model can be extended to generate spontaneous EEG signals from multiple coupled neural populations. Model parameters related to excitation, inhibition and coupling are then altered to produce epileptiform EEG signals. Results show that the qualitative behavior of the model is realistic; simulated signals resemble those recorded from different brain structures for both interictal and ictal activities. Possible exploitation of simulations in signal processing is illustrated through one example; statistical couplings between both simulated signals and real SEEG signals are estimated using nonlinear regression. Results are compared and show that, through the model, real SEEG signals can be interpreted with the aid of signal processing methods. Received: 3 January 2000 / Accepted: 24 March 2000     

Fractal dimension of electroencephalographic time series and underlying brain processes

 Fractal dimension has been proposed as a useful measure for the characterization of electrophysiological time series. This paper investigates what the pointwise dimension of electroencephalographic (EEG) time series can reveal about underlying neuronal generators. The following theoretical assumptions concerning brain function were made (i) within the cortex, strongly coupled neural assemblies exist which oscillate at certain frequencies when they are active, (ii) several such assemblies can oscillate at a time, and (iii) activity flow between assemblies is minimal. If these assumptions are made, cortical activity can be considered as the weighted sum of a finite number of oscillations (plus noise). It is shown that the correlation dimension of finite time series generated by multiple oscillators increases monotonically with the number of oscillators. Furthermore, it is shown that a reliable estimate of the pointwise dimension of the raw EEG signal can be calculated from a time series as short as a few seconds. These results indicate that (i) The pointwise dimension of the EEG allows conclusions regarding the number of independently oscillating networks in the cortex, and (ii) a reliable estimate of the pointwise dimension of the EEG is possible on the basis of short raw signals. Received: 1 September 1994/Accepted in revised form: 16 May 1995     

Nonlinear EEG analysis based on a neural mass model

The well-known neural mass model described by Lopes da Silva et al. (1976) and Zetterberg et al. (1978) is fitted to actual EEG data. This is achieved by reformulating the original set of integral equations as a continuous-discrete state space model. The local linearization approach is then used to discretize the state equation and to construct a nonlinear Kalman filter. On this basis, a maximum likelihood procedure is used for estimating the model parameters for several EEG recordings. The analysis of the noise-free differential equations of the estimated models suggests that there are two different types of alpha rhythms: those with a point attractor and others with a limit cycle attractor. These attractors are also found by means of a nonlinear time series analysis of the EEG recordings. We conclude that the Hopf bifurcation described by Zetterberg et al. (1978) is present in actual brain dynamics. Received: 11 August 1997 / Accepted in revised form: 20 April 1999     

Toluene degradation in the recycle liquid of biotrickling filters for air pollution control

Pollutant degradation in biotrickling filters for waste air treatment is generally thought to occur only in the biofilm. In two experiments with toluene degrading biotrickling filters, we show that suspended microorganisms in the recycle liquid may substantially contribute to the overall pollutant removal. Two days after reactor start up, the overall toluene elimination capacity reached a maximum of 125 g m⁻³ h⁻¹, which was twice that found during prolonged operation. High biodegradation activity in the recycle liquid fully accounted for this short-term peak of pollutant elimination. During steady-state operation, the toluene degradation in the recycle liquid was 21% of the overall elimination capacity, although the amount of suspended biomass was only 1% of the amount of immobilized biomass. The results suggest that biotrickling filter performance may be improved by selecting operating conditions allowing for the development of an actively growing suspended culture. Received: 16 June 1999 / Received revision: 17 November 1999 / Accepted: 15 December 1999     

Individual differences in brain dynamics: important implications for the calculation of event-related band power

Measures of event-related band power such as event-related desynchronization (ERD) are conventionally analyzed within fixed frequency bands, although it is known that EEG frequency varies as a function of a variety of factors. The question of how to determine these frequency bands for ERD analyses is discussed and a new method is proposed. The rationale of this new method is to adjust the frequency bands to the individual alpha frequency (IAF) for each subject and to determine the bandwidth for the alpha and theta bands as a percentage of IAF. As an example, if IAF equals 12 Hz, the widths of the alpha and theta bands are larger as compared to a subject with an IAF of, e.g., only 8 Hz. The results of an oddball paradigm show that the proposed method is superior to methods that are based on fixed frequencies and fixed bandwidths. Received: 22 July 1997 / Accepted in revised form: 22 April 1998     

Dynamic characteristics of the human resonance EEG responses to rhythmic photostimulation

The development of the resonance EEG responses of the left and right occipital areas was studied in right-handed men during prolonged (12 or 120 s) rhythmic, photostimulation with the intensity of 0.7 J and frequencies of 6, 10, and 16 Hz. Analysis of the EEG fine spectral structure was applied to compare the accumulated baseline EEG spectra and EEG spectra during photostimulation, to observe the dynamics of the short-term spectra and to detect power changes in the EEG narrow spectral band sharply coincident with the stimulation frequency. The more pronounced EEG responses to photostimulation were observed in subjects with the initially low EEG baseline, α-rhythm. Two-minute flash trains produced a substantial increase in the EEG power within the stimulation frequency with superposed oscillatory processes with different periods. These fluctuations are considered a reflection of intricate interaction between the adaptive and resonance EEG responses to the presented intermittent stimulation. Under 12-s stimulation the resonance EEG responses are steadily recorded within the first 3 s of stimulation and immediately after the flash cessation EEG power at the stimulation frequency returns to the initial level. The resonance EEG responses were more pronounced in the right hemisphere than in the left one, especially, at the stimulation frequencies of 6 and 16 Hz. With increasing the stimulation frequency, the maximum of resonance EEG responses was reached earlier. Under the stimulation frequency of 6 Hz, the maximal response was recorded 9–12 s after the beginning of flashes, at the frequencies of 10 and 16 Hz, it was recorded within 3–6 and 3 s, respectively.     

Motor Imagery EEG Spectral-Spatial Feature Optimization Using Dual-Tree Complex Wavelet and Neighbourhood Component Analysis

BackgroundFrequency band optimization improves the performance of common spatial pattern (CSP) in motor imagery (MI) tasks classification because MI-related electroencephalograms (EEGs) are highly frequency specific. Many variants of CSP algorithm divided the EEG into various sub bands and then applied CSP. However, the feature dimension of MI-EEG data increases with addition of frequency sub bands and requires efficient feature selection algorithms. The performance of CSP also depends on filtering techniques.MethodIn this study, we designed a dual tree complex wavelet transform based filter bank to filter the EEG into sub bands, instead of traditional filtering methods, which improved the spatial feature extraction efficiency. Further, after filtering EEG into different sub bands, we extracted spatial features from each sub band using CSP and optimized them by a proposed supervised learning framework based on neighbourhood component analysis (NCA). Subsequently, a support vector machine (SVM) is trained to perform classification.ResultsAn experimental study, conducted on two datasets (BCI Competition IV (Dataset 2b), and BCI competition III (Dataset IIIa)), validated the MI classification effectiveness of the proposed method in comparison with standard algorithms such as CSP, Filter bank CSP (CSP), and Discriminative FBCSP (DFBCSP). The average classification accuracy obtained by the proposed method for BCI Competition IV (Dataset 2b), and BCI Competition III (Dataset IIIa) are 84.02 ± 12.2 and 89.1 ± 7.50, respectively and found significant than that achieved by standard methods.ConclusionAchieved superior results suggest that the proposed algorithm can improve the performance of MI-based Brain-computer interface devices.     

Widely predicting specific protein functions based on protein-protein interaction data and gene expression profile

GESTs (gene expression similarity and taxonomy similarity), a gene functional prediction approach previously proposed by us, is based on gene expression similarity and concept similarity of functional classes defined in Gene Ontology (GO). In this paper, we extend this method to protein-protein interaction data by introducing several methods to filter the neighbors in protein interaction networks for a protein of unknown function(s). Unlike other conventional methods, the proposed approach automatically selects the most appropriate functional classes as specific as possible during the learning process, and calls on genes annotated to nearby classes to support the predictions to some small-sized specific classes in GO. Based on the yeast protein-protein interaction information from MIPS and a dataset of gene expression profiles, we assess the performances of our approach for predicting protein functions to “biology process” by three measures particularly designed for functional classes organized in GO. Results show that our method is powerful for widely predicting gene functions with very specific functional terms. Based on the GO database published in December 2004, we predict some proteins whose functions were unknown at that time, and some of the predictions have been confirmed by the new SGD annotation data published in April, 2006.     

Support vector machine regression for the prediction of maize hybrid performance

Accurate prediction of the phenotypical performance of untested single-cross hybrids allows for a faster genetic progress of the breeding pool at a reduced cost. We propose a prediction method based on ɛ-insensitive support vector machine regression (ɛ-SVR). A brief overview of the theoretical background of this fairly new technique and the use of specific kernel functions based on commonly applied genetic similarity measures for dominant and co-dominant markers are presented. These different marker types can be integrated into a single regression model by means of simple kernel operations. Field trial data from the grain maize breeding programme of the private company RAGT R2n are used to assess the predictive capabilities of the proposed methodology. Prediction accuracies are compared to those of one of today’s best performing prediction methods based on best linear unbiased prediction. Results on our data indicate that both methods match each other’s prediction accuracies for several combinations of marker types and traits. The ɛ-SVR framework, however, allows for a greater flexibility in combining different kinds of predictor variables.     

On the correlation dimension of optokinetic nystagmus eye movements: computational parameters, filtering, nonstationarity, and surrogate data

 We discuss the estimation of the correlation dimension of optokinetic nystagmus (OKN), a type of reflexive eye movement. Parameters of the time-delay reconstruction of the attractor are investigated, including the number of data points, the time delay, the window duration, and the duration of the signal being analyzed. Adequate values are recommended. Digital low-pass filtering causes the dimension to increase as the filter cutoff frequency decreases, in accord with a previously published prediction. The stationarity of the correlation dimension is examined; the dimension appears to decrease over the course of 120 s of continuous stimulation. Implications for the reliable estimation of the dimension are considered. Several surrogate data sets are constructed, based on both early (0–30 s) and late (100–130 s) OKN segments. Most of the surrogate data sets randomize some aspect of the original OKN, while maintaining other aspects. Dimensions are found for all surrogates and for the original OKN. Evidence is found that is consistent with some amount of deterministic and nonlinear dynamics in OKN. When this structure is randomized in the surrogate, the dimension changes or the dimension algorithm ceases to converge to a finite value. Implications for further analysis and modeling of OKN are discussed. Received: 30 August 1996/Accepted in revised form: 13 November 1996     

Topographic analysis of dimension estimates of EEG and filtered rhythms in epileptic patients with complex partial seizures

Nonlinear dynamic properties were analyzed on the EEG and filtered rhythms recorded from healthy subjects and epileptic patients with complex partial seizures. Estimates of correlation dimensions of control EEG, interictal EEG and ictal EEG were calculated. The values were demonstrated on topograms. The delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz) and gamma (30–40 Hz) components were obtained and considered as signals from the cortex. Corresponding surrogate data was produced. Firstly, the influence of sampling parameters on the calculation was tested. The dimension estimates of the signals from the frontal, temporal, parietal and occipital regions were computed and compared with the results of surrogate data. In the control subjects, the estimates between the EEG and surrogate data did not differ (P > 0.05). The interictal EEG from the frontal region and occipital region, as well as its theta component from the frontal region, and temporal region, showed obviously low dimensions (P < 0.01). The ictal EEG exhibited significantly low-dimension estimates across the scalp. All filtered rhythms from the temporal region yielded lower results than those of the surrogate data (P < 0.01). The dimension estimates of the EEG and filtered components markedly changed when the neurological state varied. For each neurological state, the dimension estimates were not uniform among the EEG and frequency components. The signal with a different frequency range and in a different neurological state showed a different dimension estimate. Furthermore, the theta and alpha components demonstrated the same estimates not only within each neurological state, but also among the different states. These results indicate that the theta and alpha components may be caused by similar dynamic processes. We conclude that the brain function underlying the ictal EEG has a simple mechanism. Several heterogeneous dynamic systems play important roles in the generation of EEG. Received: 10 December 1999 / Accepted in revised form: 8 May 2000     

Using predicted shape string to enhance the accuracy of <Emphasis Type="Bold">γ</Emphasis>-turn prediction

Numerous methods for predicting γ-turns in proteins have been developed. However, the results they generally provided are not very good, with a Matthews correlation coefficient (MCC) ≤0.18. Here, an attempt has been made to develop a method to improve the accuracy of γ-turn prediction. First, we employ the geometric mean metric as optimal criterion to evaluate the performance of support vector machine for the highly imbalanced γ-turn dataset. This metric tries to maximize both the sensitivity and the specificity while keeping them balanced. Second, a predictor to generate protein shape string by structure alignment against the protein structure database has been designed and the predicted shape string is introduced as new variable for γ-turn prediction. Based on this perception, we have developed a new method for γ-turn prediction. After training and testing the benchmark dataset of 320 non-homologous protein chains using a fivefold cross-validation technique, the present method achieves excellent performance. The overall prediction accuracy Q _total can achieve 92.2% and the MCC is 0.38, which outperform the existing γ-turn prediction methods. Our results indicate that the protein shape string is useful for predicting protein tight turns and it is reasonable to use the dihedral angle information as a variable for machine learning to predict protein folding. The dataset used in this work and the software to generate predicted shape string from structure database can be obtained from anonymous ftp site freely.     

Recognition of β-hairpin motifs in proteins by using the composite vector

A composite vector method for predicting β-hairpin motifs in proteins is proposed by combining the score of matrix, increment of diversity, the value of distance and auto-correlation information to express the information of sequence. The prediction is based on analysis of data from 3,088 non-homologous protein chains including 6,035 β-hairpin motifs and 2,738 non-β-hairpin motifs. The overall accuracy of prediction and Matthew’s correlation coefficient are 83.1% and 0.59, respectively. In addition, by using the same methods, the accuracy of 80.7% and Matthew’s correlation coefficient of 0.61 are obtained for other dataset with 2,878 non-homologous protein chains, which contains 4,884 β-hairpin motifs and 4,310 non-β-hairpin motifs. Better results are also obtained in the prediction of the β-hairpin motifs of proteins by analysis of the CASP6 dataset.     

Induction of emotional states by reading aloud texts with various emotional valences and changes in the EEG power in the β and γ frequency bands

The EEG power in the β₁, β₂, and γ frequency bands during reading according to the method of “self-regulatory utterance” was compared in subjects reading aloud emotionally neutral business texts related to an unknown field of activity, fiction texts with clear positive or negative valences or personally important autobiographic texts with similar emotional valences. Two groups of subjects participated in the study: students training to be actors (N = 22) and students with other specializations (N = 23). We observed higher values of the EEG power in the γ (30–40 Hz) and β₂ (18.5–29.5 Hz) frequency bands when comparing the states during reading of emotionally positive and emotionally negative fiction texts and personally important texts. These data are similar to our previous studies with the use of techniques that apply internal induction of positive or negative emotions without speech, in different groups of subjects. Internal induction of positive emotions was associated with an increase in the EEG power in these bands compared to the performance of an emotionally neutral task, whereas induction of negative emotions resulted in a decrease in the EEG power.     

Prediction of the β-Hairpins in Proteins Using Support Vector Machine

By using of the composite vector with increment of diversity and scoring function to express the information of sequence, a support vector machine (SVM) algorithm for predicting β-hairpin motifs is proposed. The prediction is done on a dataset of 3,088 non homologous proteins containing 6,027 β-hairpins. The overall accuracy of prediction and Matthew’s correlation coefficient are 79.9% and 0.59 for the independent testing dataset. In addition, a higher accuracy of 83.3% and Matthew’s correlation coefficient of 0.67 in the independent testing dataset are obtained on a dataset previously used by Kumar et al. (Nuclic Acid Res 33:154–159). The performance of the method is also evaluated by predicting the β-hairpins of in the CASP6 proteins, and the better results are obtained. Moreover, this method is used to predict four kinds of supersecondary structures. The overall accuracy of prediction is 64.5% for the independent testing dataset.     

A novel approach for short-term load forecasting using support vector machines

A support vector machine (SVM) modeling approach for short-term load forecasting is proposed. The SVM learning scheme is applied to the power load data, forcing the network to learn the inherent internal temporal property of power load sequence. We also study the performance when other related input variables such as temperature and humidity are considered. The performance of our proposed SVM modeling approach has been tested and compared with feed-forward neural network and cosine radial basis function neural network approaches. Numerical results show that the SVM approach yields better generalization capability and lower prediction error compared to those neural network approaches.     

A Two-Step Penalized Regression Method with Networked Predictors

Penalized regression incorporating prior dependency structure of predictors can be effective in high-dimensional data analysis (Li and Li in Bioinformatics, 24:1175–1118, 2008). Pan et al. (Biometrics, 66:474–484, 2010) proposed a penalized regression method for better outcome prediction and variable selection by smoothing parameters over a given predictor network, which can be applied to analysis of microarray data with a given gene network. In this paper, we develop two modifications to their method for further performance enhancement. First, we employ convex programming and show its improved performance over an approximate optimization algorithm implemented in their original proposal. Second, we perform bias reduction after initial variable selection through a new penalty, leading to better parameter estimates and outcome prediction. Simulations have demonstrated substantial performance improvement of the proposed modifications over the original method.     

Characteristics of EEG reactivity changes during the performance of dual tasks in healthy subjects (voluntary postural control and calculation)

Comprehensive EEG and stabilography investigation with separate and simultaneous performance of motor (voluntary postural control) and cognitive (calculation) tasks has been performed in 20 healthy subjects (22 ± 0.7 years). Specific spatial and frequency reactive changes have been found during motor task performance. These included an increase in coherence in the EEG α band for distant derivation pairs in the right hemisphere, as well as in symmetric parietal-occipital areas in both hemispheres. Cognitive task performance was accompanied by an increase in coherence for the slow bands (δ and θ) with a higher activation in the left hemisphere and frontal cortex areas. In performing the dual task, one could observe activation of spatial and frequency changes including both motor and cognitive tasks. In the dual tasks where both components were performed worse as compared to the control, reactive reorganization of EEG coherence was less pronounced than during the performance of separate tasks. A decrease in the coherence of the α₁ band in the frontal areas appeared as a zone of “conflict of interest” or interference. In dual tasks with better performance of each component as compared to the control, EEG coherence increased in each specific area, as well as in the areas of “conflict of interests.”