No abstract

Electroencephalographic (EEG) recordings are often contaminated with muscle artifacts. This disturbing myogenic activity not only strongly affects the visual analysis of EEG, but also most surely impairs the results of EEG signal processing tools such as source localization. This article focuses on the particular context of the contamination epileptic signals (interictal spikes) by muscle artifact, as EEG is a key diagnosis tool for this pathology. In this context, our aim was to compare the ability of two stochastic approaches of blind source separation, namely independent component analysis (ICA) and canonical correlation analysis (CCA), and of two deterministic approaches namely empirical mode decomposition (EMD) and wavelet transform (WT) to remove muscle artifacts from EEG signals. To quantitatively compare the performance of these four algorithms, epileptic spike-like EEG signals were simulated from two different source configurations and artificially contaminated with different levels of real EEG-recorded myogenic activity. The efficiency of CCA, ICA, EMD, and WT to correct the muscular artifact was evaluated both by calculating the normalized mean-squared error between denoised and original signals and by comparing the results of source localization obtained from artifact-free as well as noisy signals, before and after artifact correction. Tests on real data recorded in an epileptic patient are also presented. The results obtained in the context of simulations and real data show that EMD outperformed the three other algorithms for the denoising of data highly contaminated by muscular activity. For less noisy data, and when spikes arose from a single cortical source, the myogenic artifact was best corrected with CCA and ICA. Otherwise when spikes originated from two distinct sources, either EMD or ICA offered the most reliable denoising result for highly noisy data, while WT offered the better denoising result for less noisy data. These results suggest that the performance of muscle artifact correction methods strongly depend on the level of data contamination, and of the source configuration underlying EEG signals. Eventually, some insights into the numerical complexity of these four algorithms are given.

No abstract

Several studies dealing with ICA-based BCI systems have been reported. Most of them have only explored a limited number of ICA methods, mainly FastICA and INFOMAX. The aim of this paper is to help the BCI community researchers, especially those who are not familiar with ICA techniques, to choose an appropriate ICA method. For this purpose, the concept of ICA is reviewed and different measures of statistical independence are reported. Then, the application of these measures is illustrated through a brief description of the widely used algorithms in the ICA community, namely SOBI, COM2, JADE, ICAR, FastICA and INFOMAX. The implementation of these techniques in the BCI field is also explained. Finally, a comparative study of these algorithms, conducted on simulated EEG data, shows that an appropriate selection of an ICA algorithm may significantly improve the capabilities of BCI systems. Applications Neuroprosthesis Spelling Device Wheelchair, etc.

The problem of Blind Identification of linear mixtures of independent random processes is known to be related to the diagonalization of some tensors. This problem is posed here in terms of a non-conventional joint approximate diagonalization of several matrices. In fact, a congruent transform is applied to each of these matrices, the left transform being rectangular full rank, and the right one being unitary. The application in antenna signal processing is described, and suboptimal numerical algorithms are proposed.

No abstract

A number of application areas such as biomedical engineering require solving an underdetermined linear inverse problem. In such a case, it is necessary to make assumptions on the sources to restore identifiability. This problem is encountered in brain source imaging when identifying the source signals from noisy electroencephalographic or magnetoencephalographic measurements. This inverse problem has been widely studied during the last decades, giving rise to an impressive number of methods using different priors. Nevertheless, a thorough study of the latter, including especially sparse and tensorbased approaches, is still missing. In this paper, we propose i) a taxonomy of the algorithms based on methodological considerations, ii) a discussion of identifiability and convergence properties, advantages, drawbacks, and application domains of various techniques, and iii) an illustration of the performance of selected methods on identical data sets. Directions for future research in the area of biomedical imaging are eventually provided.

Electric Source Imaging (ESI) and Magnetic Source Imaging (MSI) of EEG and MEG signals are widely used to determine the origin of interictal epileptic discharges during the pre-surgical evaluation of patients with epilepsy. Epileptic discharges are detectable on EEG/MEG scalp recordings only when associated with a spatially extended cortical generator of several square centimeters, therefore it is essential to assess the ability of source localization methods to recover such spatial extent. In this study we evaluated two source localization methods that have been developed for localizing spatially extended sources using EEG/MEG data: coherent Maximum Entropy on the Mean (cMEM) and 4th order Extended Source Multiple Signal Classification (4-ExSo-MUSIC). In order to propose a fair comparison of the performances of the two methods in MEG versus EEG, this study considered realistic simulations of simultaneous EEG/MEG acquisitions taking into account an equivalent number of channels in EEG (257 electrodes) and MEG (275 sensors), involving a biophysical computational neural mass model of neuronal discharges and realistically shaped head models. cMEM and 4-ExSo-MUSIC were evaluated for their sensitivity to localize complex patterns of epileptic discharges which includes (a) different locations and spatial extents of multiple synchronous sources, and (b) propagation patterns exhibited by epileptic discharges. Performance of the source localization methods was assessed using a detection accuracy index (Area Under receiver operating characteristic Curve, AUC) and a Spatial Dispersion (SD) metric. Finally, we also presented two examples illustrating the performance of cMEM and 4-ExSo-MUSIC on clinical data recorded using high resolution EEG and MEG. When simulating single sources at different locations, both 4-ExSo-MUSIC and cMEM exhibited excellent performance (median AUC significantly larger than 0.8 for EEG and MEG), whereas, only for EEG, 4-ExSo-MUSIC showed significantly larger AUC values than cMEM. On the other hand, cMEM showed significantly lower SD values than 4-ExSo-MUSIC for both EEG and MEG. When assessing the impact of the source spatial extent, both methods provided consistent and reliable detection accuracy for a wide range of source spatial extents (source sizes ranging from 3 to 20cm for MEG and 3 to 30cm for EEG). For both EEG and MEG, 4-ExSo-MUSIC localized single source of large signal-to-noise ratio better than cMEM. In the presence of two synchronous sources, cMEM was able to distinguish well the two sources (their location and spatial extent), while 4-ExSo-MUSIC only retrieved one of them. cMEM was able to detect the spatio-temporal propagation patterns of two synchronous activities while 4-ExSo-MUSIC favored the strongest source activity. Overall, in the context of localizing sources of epileptic discharges from EEG and MEG data, 4-ExSo-MUSIC and cMEM were found accurately sensitive to the location and spatial extent of the sources, with some complementarities. Therefore, they are both eligible for a...