Saeid Sanei

Artifact removal from electroencephalograms using a hybrid BSS-SVM algorithm

Artifacts such as eye blinks and heart rhythm (ECG) cause the main interfering signals within electroencephalogram (EEG) measurements. Therefore, we propose a method for artifact removal based on exploitation of certain carefully chosen statistical features of independent components extracted from the EEGs, by fusing support vector machines (SVMs) and blind source separation (BSS). We use the second-order blind identification (SOBI) algorithm to separate the EEG into statistically independent sources and SVMs to identify the artifact components and thereby to remove such signals. The remaining independent components are remixed to reproduce the artifact-free EEGs. Objective and subjective assessment of the simulation results shows that the algorithm is successful in mitigating the interference within EEGs.

Penalty function-based joint diagonalization approach for convolutive blind separation of nonstationary sources

A new approach for convolutive blind source separation (BSS) by explicitly exploiting the second-order nonstationarity of signals and operating in the frequency domain is proposed. The algorithm accommodates a penalty function within the cross-power spectrum-based cost function and thereby converts the separation problem into a joint diagonalization problem with unconstrained optimization. This leads to a new member of the family of joint diagonalization criteria and a modification of the search direction of the gradient-based descent algorithm. Using this approach, not only can the degenerate solution induced by a null unmixing matrix and the effect of large errors within the elements of covariance matrices at low-frequency bins be automatically removed, but in addition, a unifying view to joint diagonalization with unitary or nonunitary constraint is provided. Numerical experiments are presented to verify the performance of the new method, which show that a suitable penalty function may lead the algorithm to a faster convergence and a better performance for the separation of convolved speech signals, in particular, in terms of shape preservation and amplitude ambiguity reduction, as compared with the conventional second-order based algorithms for convolutive mixtures that exploit signal nonstationarity.

An adaptive singular spectrum analysis approach to murmur detection from heart sounds

Murmur is the result of various heart abnormalities. A new robust approach for separation of murmur from heart sound has been suggested in this article. Singular spectrum analysis (SSA) has been adapted to the changes in the statistical properties of the data and effectively used for detection of murmur from single-channel heart sound (HS) signals. Incorporating a cleverly selected a priori within the SSA reconstruction process, results in an accurate separation of normal HS from the murmur segment. Another contribution of this work is selection of the correct subspace of the desired signal component automatically. In addition, the subspace size can be identified iteratively. A number of HS signals with murmur have been processed using the proposed adaptive SSA (ASSA) technique and the results have been quantified both objectively and subjectively.

A New Adaptive Line Enhancer Based on Singular Spectrum Analysis

A new ALE-based on singular spectrum analysis (SSA) is proposed here. In this approach in the reconstruction stage of SSA the eigentriples are adaptively selected using the delayed version of the data. Unlike for the conventional ALE where order statistics are taken into account, here full eigen-spectrum of the embedding matrix is exploited. Consequently, the system works for non-Gaussian noise and wideband periodic signals. The performance of the system is demonstrated for synthetic as well as real signals and compared with those of traditional ALE.

Localizing Heart Sounds in Respiratory Signals Using Singular Spectrum Analysis

Respiratory sounds are always contaminated by heart sound interference. An essential preprocessing step in some of the heart sound cancellation methods is localizing primary heart sound components. Singular spectrum analysis (SSA), a powerful time series analysis technique, is used in this paper. Despite the frequency overlap of the heart and lung sound components, two different trends in the eigenvalue spectra are recognizable, which leads to find a subspace that contains more information about the underlying heart sound. Artificially mixed and real respiratory signals are used for evaluating the performance of the method. Selecting the appropriate length for the SSA window results in good decomposition quality and low computational cost for the algorithm. The results of the proposed method are compared with those of well-established methods, which use the wavelet transform and entropy of the signal to detect the heart sound components. The proposed method outperforms the wavelet-based method in terms of false detection and also correlation with the underlying heart sounds. Performance of the proposed method is slightly better than that of the entropy-based method. Moreover, the execution time of the former is significantly lower than that of the latter.

A gradient-based alternating minimization approach for optimization of the measurement matrix in compressive sensing

In this paper the problem of optimization of the measurement matrix in compressive (also called compressed) sensing framework is addressed. In compressed sensing a measurement matrix that has a small coherence with the sparsifying dictionary (or basis) is of interest. Random measurement matrices have been used so far since they present small coherence with almost any sparsifying dictionary. However, it has been recently shown that optimizing the measurement matrix toward decreasing the coherence is possible and can improve the performance. Based on this conclusion, we propose here an alternating minimization approach for this purpose which is a variant of Grassmannian frame design modified by a gradient-based technique. The objective is to optimize an initially random measurement matrix to a matrix which presents a smaller coherence than the initial one. We established several experiments to measure the performance of the proposed method and compare it with those of the existing approaches. The results are encouraging and indicate improved reconstruction quality, when utilizing the proposed method.

A unified approach to sparse signal processing

A unified view of the area of sparse signal processing is presented in tutorial form by bringing together various fields in which the property of sparsity has been successfully exploited. For each of these fields, various algorithms and techniques, which have been developed to leverage sparsity, are described succinctly. The common potential benefits of significant reduction in sampling rate and processing manipulations through sparse signal processing are revealed. The key application domains of sparse signal processing are sampling, coding, spectral estimation, array processing, component analysis, and multipath channel estimation. In terms of the sampling process and reconstruction algorithms, linkages are made with random sampling, compressed sensing, and rate of innovation. The redundancy introduced by channel coding in finite and real Galois fields is then related to over-sampling with similar reconstruction algorithms. The error locator polynomial (ELP) and iterative methods are shown to work quite effectively for both sampling and coding applications. The methods of Prony, Pisarenko, and MUltiple SIgnal Classification (MUSIC) are next shown to be targeted at analyzing signals with sparse frequency domain representations. Specifically, the relations of the approach of Prony to an annihilating filter in rate of innovation and ELP in coding are emphasized; the Pisarenko and MUSIC methods are further improvements of the Prony method under noisy environments. The iterative methods developed for sampling and coding applications are shown to be powerful tools in spectral estimation. Such narrowband spectral estimation is then related to multi-source location and direction of arrival estimation in array processing. Sparsity in unobservable source signals is also shown to facilitate source separation in sparse component analysis; the algorithms developed in this area such as linear programming and matching pursuit are also widely used in compressed sensing. Finally, the multipath channel estimation problem is shown to have a sparse formulation; algorithms similar to sampling and coding are used to estimate typical multicarrier communication channels.

Epileptic seizure predictability from scalp EEG incorporating constrained blind source separation

Most of the methods for prediction of epilepsy recently reported in the literature are based on the evaluation of chaotic behavior of intracranial electroencephalographic (EEG) recordings. These recordings require intensive surgical operations to implant the electrodes within the brain which are hazardous to the patient. Here, we have developed a novel approach to quantify the dynamical changes of the brain using the scalp EEG. The scalp signals are preprocessed by means of an effective block-based blind source separation (BSS) technique to separate the underlying sources within the brain. The algorithm significantly removes the effect of eye blinking artifacts. An overlap window procedure has been incorporated in order to mitigate the inherent permutation problem of BSS and maintain the continuity of the estimated sources. Chaotic behavior of the underlying sources has then been evaluated by measuring the largest Lyapunov exponent. For our experiments, we provided twenty sets of simultaneous intracranial and scalp EEG recordings from twenty patients. The above recordings have been compared. Similar results were obtained when the intracranial electrodes recorded the electrical activity of the epileptic focus. Our preliminary results show a great improvement when the epileptic focus is not captured by the intracranial electrodes

Fast and incoherent dictionary learning algorithms with application to fMRI

In this paper, the problem of dictionary learning and its analogy to source separation is addressed. First, we extend the well-known method of K-SVD to incoherent K-SVD, to enforce the algorithm to achieve an incoherent dictionary. Second, a fast dictionary learning algorithm based on steepest descent method is proposed. The main advantage of this method is high speed since both coefficients and dictionary elements are updated simultaneously rather than column-by-column. Finally, we apply the proposed methods to both synthetic and real functional magnetic resonance imaging data for the detection of activated regions in the brain. The results of our experiments confirm the effectiveness of the proposed ideas. In addition, we compare the quality of results and empirically prove the superiority of the proposed dictionary learning methods over the conventional algorithms.

Blind Separation of Image Sources via Adaptive Dictionary Learning

Sparsity has been shown to be very useful in source separation of multichannel observations. However, in most cases, the sources of interest are not sparse in their current domain and one needs to sparsify them using a known transform or dictionary. If such a priori about the underlying sparse domain of the sources is not available, then the current algorithms will fail to successfully recover the sources. In this paper, we address this problem and attempt to give a solution via fusing the dictionary learning into the source separation. We first define a cost function based on this idea and propose an extension of the denoising method in the work of Elad and Aharon to minimize it. Due to impracticality of such direct extension, we then propose a feasible approach. In the proposed hierarchical method, a local dictionary is adaptively learned for each source along with separation. This process improves the quality of source separation even in noisy situations. In another part of this paper, we explore the possibility of adding global priors to the proposed method. The results of our experiments are promising and confirm the strength of the proposed approach.