Wassim El Falou

Evaluation of driver discomfort during long-duration car driving ☆

The evolution of indices of fatigue, discomfort, and performance of subjects seated for long duration (150 min) in car seats were studied (n=11). Four experimental configurations were used: with and without vibration for two seats (U, uncomfortable; C, comfortable). Surface electromyography (SEMG) data were recorded bilaterally from cervical erector spinae and external oblique muscles. Discomfort increased significantly during the trial, regardless of the experimental condition (p<0.05). Performance was significantly worse for seat U with vibration (p<0.05). The median frequency of SEMG signals did not change between experimental conditions or across time. It would appear that, either the level of discomfort experienced was insufficient to change either performance or SEMG measures, or that the large parameter estimation variance of the SEMG signals might have masked any underlying spectral change. Further refinement of the SEMG signal processing methodology may be necessary to be able to detect fatigue of postural muscles.

EEGNET: An Open Source Tool for Analyzing and Visualizing M/EEG Connectome.

The brain is a large-scale complex network often referred to as the “connectome”. Exploring the dynamic behavior of the connectome is a challenging issue as both excellent time and space resolution is required. In this context Magneto/Electroencephalography (M/EEG) are effective neuroimaging techniques allowing for analysis of the dynamics of functional brain networks at scalp level and/or at reconstructed sources. However, a tool that can cover all the processing steps of identifying brain networks from M/EEG data is still missing. In this paper, we report a novel software package, called EEGNET, running under MATLAB (Math works, inc), and allowing for analysis and visualization of functional brain networks from M/EEG recordings. EEGNET is developed to analyze networks either at the level of scalp electrodes or at the level of reconstructed cortical sources. It includes i) Basic steps in preprocessing M/EEG signals, ii) the solution of the inverse problem to localize / reconstruct the cortical sources, iii) the computation of functional connectivity among signals collected at surface electrodes or/and time courses of reconstructed sources and iv) the computation of the network measures based on graph theory analysis. EEGNET is the unique tool that combines the M/EEG functional connectivity analysis and the computation of network measures derived from the graph theory. The first version of EEGNET is easy to use, flexible and user friendly. EEGNET is an open source tool and can be freely downloaded from this webpage: https://sites.google.com/site/eegnetworks/.

Automatic Threshold Determination for a Local Approach of Change Detection in Long-Term Signal Recordings

CUSUM (cumulative sum) is a well-known method that can be used to detect changes in a signal when the parameters of this signal are known. This paper presents an adaptation of the CUSUM-based change detection algorithms to long-term signal recordings where the various hypotheses contained in the signal are unknown. The starting point of the work was the dynamic cumulative sum (DCS) algorithm, previously developed for application to long-term electromyography (EMG) recordings. DCS has been improved in two ways. The first was a new procedure to estimate the distribution parameters to ensure the respect of the detectability property. The second was the definition of two separate, automatically determined thresholds. One of them (lower threshold) acted to stop the estimation process, the other one (upper threshold) was applied to the detection function. The automatic determination of the thresholds was based on the Kullback-Leibler distance which gives information about the distance between the detected segments (events). Tests on simulated data demonstrated the efficiency of these improvements of the DCS algorithm.

Hardware implementation of a fault-tolerant Hopfield Neural Network on FPGAs

This letter presents an FPGA implementation of a fault-tolerant Hopfield Neural Network (HNN). The robustness of this circuit against Single Event Upsets (SEUs) and Single Event Transients (SETs) has been evaluated. Results show the fault tolerance of the proposed design, compared to a previous non-fault-tolerant implementation and a solution based on triple modular redundancy (TMR) of a standard HNN design.

Tracking System using GPS, vision and 3D virtual Model

In order to improve the vehicle tracking quality in the cities and especially in urban area, the following article handles the correspondence between real and virtual images to find the closest virtual image to the real one. Real image are extracted from cameras equipped by a GPS system, together installed in the vehicle. Virtual images are extracted using a GPS from a database managed by 3D geographical information system (3D- GIS). It is known that GPS cannot give accurately the coordinates of a vehicle so it is necessary to use other kind of information using embedded sensors like camera. A way to compute a position using vision is to find the closest image in a 3D cartographical database which corresponds to the real one seen by the camera. Two methods are developed and tested with real data : the first method uses the Hough transform where each line corresponds to a point in the polar coordinate then we compare the image transformations. The second method is based on the Ransac fitting homography method. This method based on taking the two images real and virtual image, find the corners of each image using a harris corner detector, use the maximally correlated points to connect them, robustly fits a homography to a set of putatively matched image points, find the number of putatively matched image points that are called inliers and the greatest the number of inliers the closer is the virtual image to a real one. It uses homography, harris corner detector, and correlation functions. Results with real data are presented to illustrate performance of developed method.

Efficient depth map compression exploiting correlation with texture data in multiresolution predictive image coders

New 3D applications such as 3DTV and FVV require not only a large amount of data, but also high-quality visual rendering. Based on one or several depth maps, intermediate views can be synthesized using a depth image-based rendering technique. Many compression schemes have been proposed for texture-plus-depth data, but the exploitation of the correlation between the two representations in enhancing compression performances is still an open research issue. In this paper, we present a novel compression scheme that aims at improving the depth coding using a joint depth/texture coding scheme. This method is an extension of the LAR (Locally Adaptive Resolution) codec, initially designed for 2D images. The LAR coding framework provides a lot of functionalities such as lossy/lossless compression, low complexity, resolution and quality scalability and quality control. Experimental results address both lossless and lossy compression aspects, considering some state of the art techniques in the two domains (JPEGLS, JPEGXR). Subjective results on the intermediate view synthesis after depth map coding show that the proposed method significantly improves the visual quality.

Classification of pregnancy and labor contractions using a graph theory based analysis.

In this paper, we propose a new framework to characterize the electrohysterographic (EHG) signals recorded during pregnancy and labor. The approach is based on the analysis of the propagation of the uterine electrical activity. The processing pipeline includes i) the estimation of the statistical dependencies between the different recorded EHG signals, ii) the characterization of the obtained connectivity matrices using network measures and iii) the use of these measures in clinical application: the classification between pregnancy and labor. Due to its robustness to volume conductor, we used the imaginary part of coherence in order to produce the connectivity matrix which is then transformed into a graph. We evaluate the performance of several graph measures. We also compare the results with the parameter mostly used in the literature: the peak frequency combined with the propagation velocity (PV +PF). Our results show that the use of the network measures is a promising tool to classify labor and pregnancy contractions with a small superiority of the graph strength over PV+PF.

A method and an automated tool to perform SET fault-injection on HDL-based designs

A fully automated fault-injection method is presented. It deals with transient faults resulting from the impact of energetic particles and it can be applied early at design phase, on any circuit for which the register transfer level model is available. Results issued from its application to an Artificial Neural Network benchmark application put in evidence the accuracy of the studied method to predict error rates due to transient faults generated by the radiation environment.

Pregnancy monitoring using graph theory based analysis

Monitoring pregnancy using noninvasive recordings of uterine contractions is still an unsolved issue. Here, we propose a new way to tackle this problem using the electrohysterographic (EHG) signals recorded during pregnancy and labor. The new approach is based on the analysis of the propagation of the uterine electrical activity. The proposed pipeline includes i) the computation of the statistical dependencies between the multichannel (4 × 4 matrix) EHG signals, ii) the characterization of the connectivity matrices using network measures (graph-theory based analysis) and iii) the use of these measures in pregnancy monitoring. Due to its robustness to volume conduction, we used the imaginary part of coherence function to create the connectivity matrices transformed then into graphs. The method is evaluated on a dataset of EHG signals to track the correlation between uterine signals with weeks before labor. The results show a difference in the graph densities from pregnancy to labor. We speculate that the network based analysis is a very promising tool for pregnancy monitoring.

A scalp-EEG network-based analysis of Alzheimer's disease patients at rest

Most brain disorders including Alzheimers disease (AD) are related to alterations in the normal brain network organization and function. Exploring these network alterations using non-invasive and easy to use technique is a topic of great interest. In this paper, we collected EEG resting-state data from AD patients and healthy control subjects. Functional connectivity between scalp EEG signals was quantified using the phase locking value (PLV) for 6 frequency bands, θ (4–8 Hz), α1(8–10 Hz), α2(10–13 Hz), ß(13–30 Hz), γ(30–45 Hz), and broad band (0.2–45 Hz). To assess the differences in network properties, graph-theoretical analysis was performed. AD patients showed decrease of mean connectivity, average clustering and global efficiency in the lower alpha band. Positive correlation between the cognitive score and the extracted graph measures was obtained, suggesting that EEG could be a promising technique to derive new biomarkers of AD diagnosis.