Antnio Teixeira

On the use of clustering and local singular spectrum analysis to remove ocular artifacts from electroencephalograms

We present a method based on singular spectrum analysis to remove ocular artifacts (EOG) from an electroencephalogram (EEC). After embedding the EEG signals in a feature space of time-delayed coordinates, feature vectors are clustered and the principal components (PCs) are computed locally within each cluster. Then we assume that the EOG artifact is associated with the PCs belonging to the largest eigenvalues. We incorporate a minimum description length (IMDL) criterion to estimate the number of eigenvectors needed to represent the EOG artifact faithfully. The EOG signal thus extracted is then subtracted from the original EEG signal to obtain the corrected EEG signal we are interested in.

Blind source separation using time-delayed signals

In this work a modified version of AMUSE, called dAMUSE, is proposed. The main modification consists in increasing the dimension of the data vectors by joining delayed versions of the observed mixed signals. With the new data a matrix pencil is computed and its generalized eigendecomposition is performed as in AMUSE. We will show that in this case the output (or independent) signals are filtered versions of the source signals. Some numerical simulations using artificially mixed signals as well as biological data (RR and QT intervals of Electrocardiogram) are presented.

Subspace techniques to remove artifacts from EEG: A quantitative analysis

In this work we discuss and apply projective subspace techniques to both multichannel as well as single channel recordings. The single-channel approach is based on singular spectrum analysis(SSA) and the multichannel approach uses the extended infomax algorithm which is implemented in the open-source toolbox EEGLAB. Both approaches will be evaluated using artificial mixtures of a set of selected EEG signals. The latter were selected visually to contain as the dominant activity one of the characteristic bands of an electroencephalogram (EEG). The evaluation is performed both in the time and frequency domain by using correlation coefficients and coherence function, respectively.

Greedy Kernel PCA Applied to Single-Channel EEG Recordings

In this work, we propose the correction of univariate, single channel EEGs using a kernel technique. The EEG signal is embedded in its time-delayed coordinates obtaining a multivariate signal. A kernel subspace technique is used for denoising and artefact extraction. The proposed kernel method follows a greedy approach to use a reduced data set to compute a new basis onto which to project the mapped data in feature space. The pre-image of the reconstructed multivariate signal is computed and the embedding is reverted. The resultant signal is the high amplitude artifact which must be subtracted from the original signal to obtain a corrected version of the underlying signal.

On the Use of KPCA to Extract Artifacts in One-Dimensional Biomedical Signals

Kernel principal component analysis (KPCA) is a nonlinear projective technique that can be applied to decompose multi-dimensional signals and extract informative features as well as reduce any noise contributions. In this work we extend KPCA to extract and remove artifact-related contributions as well as noise from one-dimensional signal recordings. We introduce an embedding step which transforms the one-dimensional signal into a multi-dimensional vector. The latter is decomposed in feature space to extract artifact related contaminations. We further address the pre- image problem and propose an initialization procedure to the fixed-point algorithm which renders it more efficient. Finally we apply KPCA to extract dominant Electrooculogram (EOG) artifacts contaminating Electroencephalogram (EEG) recordings in a frontal channel.

Denoising using local ICA and kernel-PCA

We present a denoising algorithm for enhancing noisy signals based on local independent component analysis (ICA). This is done by applying ICA to the signal in localized delayed coordinates. The components resembling the signals can be detected by various criteria depending on the nature of the signal. Estimators of kurtosis or the variance of the autocorrelation have been considered. The algorithm proposed can favorably be applied to the problem of denoising multidimensional data like images or fMRI data sets. In comparison to denoising algorithms using wavelets, Wiener filters and kernel PCA the local PCA and ICA algorithms perform considerably better. We provide applications of the algorithm to images and the analysis of protein NMR spectra.

Kernel-PCA denoising of artifact-free protein NMR spectra

Multidimensional /sup 1/H NMR spectra of biomolecules dissolved in light water are contaminated by an intense water artifact. Generalized eigenvalue decomposition methods using congruent matrix pencils are used to separate the water artefact from the protein spectra. Due to the statistical separation process, however, noise is introduced into the reconstructed spectra. Hence Kernel-based denoising techniques are discussed to obtain noise- and artifact-free 2D NOESY NMR spectra of proteins.

A method to extract articulatory parameters from the speech signal using neural networks

We present a method that uses artificial neural networks for acoustic to articulatory mapping. An assembly of Kohonen (1982) neural nets is used, in the first stage a network maps cepstral values, each neuron contains a subnet in a second stage that maps the articulatory space. The method allows both the acoustic to articulatory mapping, ensuring smooth varying vocal tract shapes, and the study of the nonuniqueness problem.

ERP correlates of error processing during performance on the Halstead Category Test.

The Halstead Category Test (HCT) is a neuropsychological test that measures a persons ability to formulate and apply abstract principles. Performance must be adjusted based on feedback after each trial and errors are common until the underlying rules are discovered. Event-related potential (ERP) studies associated with the HCT are lacking. This paper demonstrates the use of a methodology inspired on Singular Spectrum Analysis (SSA) applied to EEG signals, to remove high amplitude ocular and movement artifacts during performance on the test. This filtering technique introduces no phase or latency distortions, with minimum loss of relevant EEG information. Importantly, the test was applied in its original clinical format, without introducing adaptations to ERP recordings. After signal treatment, the feedback-related negativity (FRN) wave, which is related to error-processing, was identified. This component peaked around 250ms, after feedback, in fronto-central electrodes. As expected, errors elicited more negative amplitudes than correct responses. Results are discussed in terms of the increased clinical potential that coupling ERP information with behavioral performance data can bring to the specificity of the HCT in diagnosing different types of impairment in frontal brain function.