Dania Gutiérrez

Publications

• 1.14

  Impact points

  EEG data classification through signal spatial redistribution and optimized linear discriminants.

  David Gutiérrez, Diana I Escalona-Vargas

  Computer methods and programs in biomedicine. 07/2009;

  This paper presents a preprocessing technique for improving the classification of electroencephalographic (EEG) data in brain-computer interfaces (BCI) for the case of realistic measuring conditions, such as low signal-to-noise ratio (SNR), reduced number of measuring electrodes, and reduced amount ... [more] This paper presents a preprocessing technique for improving the classification of electroencephalographic (EEG) data in brain-computer interfaces (BCI) for the case of realistic measuring conditions, such as low signal-to-noise ratio (SNR), reduced number of measuring electrodes, and reduced amount of data used to train the classifier. The proposed method is based on a linear minimum mean squared error (LMMSE) spatial filter specifically designed to improve the SNR of the signals before being classified. The design parameters of the spatial filter are obtained through an optimized version of Fisher's linear discriminant (FLD) whose area under the receiver operating characteristics (ROC) curve is maximized. The combination of the spatial filter and the optimized FLD increases the SNR and changes the spatial distribution of the measured signals. As a result, the signals can be more easily discriminated by means of a simple sign detector or threshold-based classifier. A series of experiments on simulated EEG data compare the performance of the proposed classification scheme to the performance of the Mahalanobis distance-based classifier, which is widely used in BCI systems. Numerical results show that the proposed preprocessing technique enhances the classifier's performance even for low SNR conditions and few measurements, while the Mahalanobis classifier is not reliable under such realistic operating conditions. Furthermore, real EEG data from a self-paced key typing experiment is used to demonstrate the applicability of the preprocessing technique. The proposed method has the potential of improving the efficiency of real-life BCI systems, as well as reducing the computational complexity associated with their implementation.
• 2.15

  Impact points

  Array Response Kernels for EEG and MEG in Multilayer Ellipsoidal Geometry

  D. Gutierrez, A. Nehorai

  Biomedical Engineering, IEEE Transactions on. 04/2008;

  We present forward modeling solutions in the form of array response kernels for electroencephalography (EEG) and magnetoencephalography (MEG), assuming that a multilayer ellipsoidal geometry approximates the anatomy of the head and a dipole current models the source. The use of an ellipsoidal geomet... [more] We present forward modeling solutions in the form of array response kernels for electroencephalography (EEG) and magnetoencephalography (MEG), assuming that a multilayer ellipsoidal geometry approximates the anatomy of the head and a dipole current models the source. The use of an ellipsoidal geometry is useful in cases for which incorporating the anisotropy of the head is important but a better model cannot be defined. The structure of our forward solutions facilitates the analysis of the inverse problem by factoring the lead field into a product of the current dipole source and a kernel containing the information corresponding to the head geometry and location of the source and sensors. This factorization allows the inverse problem to be approached as an explicit function of just the location parameters, which reduces the complexity of the estimation solution search. Our forward solutions have the potential of facilitating the solution of the inverse problem, as they provide algebraic representations suitable for numerical implementation. The applicability of our models is illustrated with numerical examples on real EEG/MEG data of N20 responses. Our results show that the residual data after modeling the N20 response using a dipole for the source and an ellipsoidal geometry for the head is in average lower than the residual remaining when a spherical geometry is used for the same estimated dipole.
• 2.15

  Impact points

  Performance analysis of reduced-rank beamformers for estimating dipole source signals using EEG/MEG

  D. Gutierrez, A. Nehorai, A. Dogandzic

  Biomedical Engineering, IEEE Transactions on. 06/2006;

  We study the performance of various beamformers for estimating a current dipole source at a known location using electroencephalography (EEG) and magnetoencephalography(MEG). We present our beamformers in the form of the generalized sidelobe canceler (GSC). Under this structure, the beamformer can b... [more] We study the performance of various beamformers for estimating a current dipole source at a known location using electroencephalography (EEG) and magnetoencephalography(MEG). We present our beamformers in the form of the generalized sidelobe canceler (GSC). Under this structure, the beamformer can be solved by finding a filter that achieves the minimum mean-squared error (MMSE) between the mainbeam response and filtered observed signal. We express the MMSE as a function of the filter's rank and use it as a criterion to evaluate the performance of the beamformers. We do not make any assumptions on the rank of the interference-plus-noise covariance matrix. Instead, we treat it as low-rank and derive a general expression for the MMSE. We present numerical examples to compare the MSE performance of beamformers commonly studied in the literature: principal components (PCs),cross-spectral metrics (CSMs), and eigencanceler (EIG) beamformers. Our results show that good estimates of the dipole source signals can be achieved using reduced-rank beamformers even for low signal-to-noise ratio (SNR) values
• 2.15

  Impact points

  Estimating brain conductivities and dipole source signals with EEG arrays

  D. Gutierrez, A. Nehorai, C.H. Muravchik

  Biomedical Engineering, IEEE Transactions on. 01/2005;

  Techniques based on electroencephalography (EEG) measure the electric potentials on the scalp and process them to infer the location, distribution, and intensity of underlying neural activity. Accuracy in estimating these parameters is highly sensitive to uncertainty in the conductivities of the hea... [more] Techniques based on electroencephalography (EEG) measure the electric potentials on the scalp and process them to infer the location, distribution, and intensity of underlying neural activity. Accuracy in estimating these parameters is highly sensitive to uncertainty in the conductivities of the head tissues. Furthermore, dissimilarities among individuals are ignored when standardized values are used. In this paper, we apply the maximum-likelihood and maximum a posteriori (MAP) techniques to simultaneously estimate the layer conductivity ratios and source signal using EEG data. We use the classical 4-sphere model to approximate the head geometry, and assume a known dipole source position. The accuracy of our estimates is evaluated by comparing their standard deviations with the Crame´r-Rao bound (CRB). The applicability of these techniques is illustrated with numerical examples on simulated EEG data. Our results show that the estimates have low bias and attain the CRB for sufficiently large number of experiments. We also present numerical examples evaluating the sensitivity to imprecise assumptions on the source position and skull thickness. Finally, we propose extensions to the case of unknown source position and present examples for real data.

• Ellipsoid Head Model for Fetal Magnetoencephalography: Forward and Inverse Solutions

  D. Gutiérrez, A. Nehorai, H. Preissl

  Physics in Medicine and Biology. 01/2005; 50:2141-2157.

• MEG source estimation in the presence of low-rank interference using cross-spectral metrics

  D. Gutierrez, A. Nehorai, A. Dogandzic

  Engineering in Medicine and Biology Society, 2004. IEMBS '04. 26th Annual International Conference of the IEEE; 10/2004

  We estimate a source current dipole at a known location in the presence of low-rank interference using magnetoencephalography (MEG). We present a space-time processor for MEG data based on the generalized sidelobe canceler (GSC). We extend the classical vector beamformer to a matrix structure withou... [more] We estimate a source current dipole at a known location in the presence of low-rank interference using magnetoencephalography (MEG). We present a space-time processor for MEG data based on the generalized sidelobe canceler (GSC). We extend the classical vector beamformer to a matrix structure without making any assumptions on the rank of the covariance matrix of noise and interference, or constraint matrices. Furthermore, we define the cross-spectral metrics (CSM) in their most general form. The CSM method is known to approximate the performance of the matched filter for the case of unknown covariance matrix. In our case, the CSM also allows to reduce the complexity of the filtering problem without significant loss of performance in the signal-to-interference-plus-noise ratio (SINR). Our results show that good estimates of the dipole sources can be achieved by only using a few eigenvalues, namely, those corresponding to the largest CSM.

• Using eigenstructure decompositions of time-varying autoregressions in common spatial patterns-based EEG signal classification

  David Gutiérrez, Rocio Salazar-Varas

  Biomedical Signal Processing and Control.