Fast feature selection to compare broadband with narrowband phase synchronization in brain-computer interfaces.

OBJECTIVE: Brain-computer interface (BCI) research aims at developing communication devices for the motor disabled. Such devices are not driven by muscle activity, but by brain activity recorded during different mental tasks. We present here the comparison of phase synchronization and power spectral density (PSD) features, computed from broadband and narrowband filtered EEG signals and their ability to discriminate three mental tasks. METHODS: EEG signals were recorded from five subjects while performing left and right hand movement imagination and word generation. We applied a modified Fast Correlation Based Filter (FCBF) [9] for the purpose of feature selection. RESULTS: We found that the features were selected from electrode signals corresponding to neurophysiological evidence, i.e. electrodes lying over the motor cortex. PSD and phase locking value (PLV) features were more discriminative when computed from narrowband (8-12 Hz) and broadband (8-30 Hz) filtered signals respectively. CONCLUSIONS: The generalization performance is as good as the one obtained with SVM-rfe, but this algorithm is faster and selects fewer features. These properties may make FCBF a valuable tool for further improvement of BCIs.

Accurate walking and running speed estimation using wrist inertial data.

In this work, we present an accelerometry-based device for robust running speed estimation integrated into a watch-like device. The estimation is based on inertial data processing, which consists in applying a leg-and-arm dynamic motion model to 3D accelerometer signals. This motion model requires a calibration procedure that can be done either on a known distance or on a constant speed period. The protocol includes walking and running speeds between 1.8km/h and 19.8km/h. Preliminary results based on eleven subjects are characterized by unbiased estimations with 2(nd) and 3(rd) quartiles of the relative error dispersion in the interval ±5%. These results are comparable to accuracies obtained with classical foot pod devices.

Physical activity profiling: Activity-specific step counting and energy expenditure models using 3D wrist acceleration.

In this paper, we present the evaluation of a new physical activity profiling system embedded in a wrist-located device. We propose a step counting and an energy expenditure (EE) method, and evaluate their accuracy against gold standard references. To this end, we used an actimetry sensor on the waist and an indirect calorimetry monitoring device on a population of 13 subjects to obtain step count and metabolic equivalent task (kcal/kg/h) referenced values. The subjects followed a protocol that spanned a given set of activities (lying, standing, walking, running) at a wide range of intensities. The performance of the EE model was characterized by a root-mean-square error (RMSE) of 1.22±0.34kcal/min, and step-count model at regular walking/running speeds by 0.71±0.06step/10sec.

Nonlinear nonstationary Wiener model of infant EEG seizures.

This paper presents the estimation of a nonstationary nonlinear model of seizures in infants based on parallel Wiener structures. The model comprises two parts and is partly derived from the Roessgen et al. seizure model. The first part consists of a nonlinear Wiener model of the pure background activity, and the second part in a nonlinear Wiener model of the pure seizure activity with a time-varying deterministic input signal. The two parts are then combined in a parallel structure. The Wiener model consists of an autoregressive moving average filter followed by a nonlinear shaping function to take into account the non-Gaussian statistical behavior of the data. Model estimation was performed on 64 infants of whom four showed signs of clinical and electrical seizures. Model validation is performed using time-frequency-based entropy distance and shows an averaged improvement of 50% in modeling performance compared with the Roessgen model.

Observer of autonomic cardiac outflow based on blind source separation of ECG parameters.

We present a novel method which provides an observer of the autonomic cardiac outflow using heartbeat intervals (RR) and QT intervals. The model of the observer is inferred from qualitative physiological knowledge. It consists in a problem of blind source separation of noisy mixtures which is resolved by a simple and robust algorithm. The robustness of the algorithm has been assessed by numerical simulations in adverse noisy environments. In clinical applications, we have validated the observer on subjects exposed to experimental conditions known to elicit sympathetic or parasympathetic response.

Observer of the human cardiac sympathetic nerve activity using noncausal blind source separation.

We present a novel method for the blind reconstruction of the cardiac sympathetic nerve activity (CSNA) in the low-frequency (LF) band (0.04-0.15 Hz) using only heart rate and arterial blood pressure. The originality of the method consists in the application of blind source separation techniques to obtain an observer of CSNA. We show how this observer can be deduced from a linear model of the cardiovascular system by introduction of the fundamental assumptions about the independence of the cardiac sympathetic an parasympathetic outflow. In cardiovascular applications, the reliability of the observer has been assessed by verification of the fundamental assumption for the given data. A primer qualitative validation has been performed using the muscle sympathetic nerve activity as an indirect indicator of CSNA. Very satisfying and promising results have been obtained. Moreover, we have performed quantitative validations of the observer in various experimental conditions known to elicit selectively cardiac sympathetic or parasympathetic response. The experimental conditions include a supine-to-60 degrees tilt test, indirect sympathetic stimulation/inhibition by medication, and sympathetic stimulation by isometric handgrip. We show that the observer allows to highlight changing levels of the cardiac sympathetic activity in the LF band in all these experimental conditions.

Time-varying statistical dimension analysis with application to newborn scalp EEG seizure signals.

A new approach to the analysis of nonstationary possibly nonlinear time series is presented. It is based on an adaptive autocovariance eigenspectrum computation known as APEX together with the Rissanen's Minimum Description Length criterion for the selection of the most relevant eigenvalues. A new concept of time-varying instantaneous statistical dimension is introduced. The motivation for this new approach is the analysis of newborn electroencephalogram for which nonstationary is an inherent property. The proposed algorithm and new dimension are first assessed on synthetic data. Then, newborn scalp EEG data are analyzed using the proposed scheme. Transitions between different brain states are shown to occur on a baby having electrical and clinical seizures.

Exponential-type distribution of human muscle sympathetic nerve activity results in an automatic quantification method.

A new burst counting method based on a subject invariant characteristic demonstrates the limits of the actual automatic based methods. The exponential behaviour of the counted bursts in function of a variable threshold highlights a scaling property of the muscle sympathetic nerve activity. From experimental single unit recording results, we deduce the exponential-type (gamma) distribution of instantaneous spiking frequency within multi-unit recordings. We show that integrated muscle sympathetic nerve discharges must be gamma distributed with parameters proportional to the number of neurons in the recording pool and to the integration window width.

Subband modeling of the human cardiovascular system: new insights into cardiovascular regulation.

We present a new approach to cardiovascular analysis based on a well-known signal processing technique, namely, the frequency subband decomposition. The subbands are chosen in accordance with physiological standards: (1) 0-0.04 Hz, (2) 0.04-0.15 Hz, (3) 0.15-0.4 Hz. It is shown that such a pre-processing drastically improves the accuracy of the analysis and introduces a new direction in the understanding of the relationships between cardiovascular signals.