Brain Signal Processing Research Articles

Comparison of pre-amplifier topologies for use in brain-computer interface with capacitively-coupled EEG electrodes

Purpose Interest in brain computer interfaces (BCI) has recently increased due to the need for quality of life technologies for disabled people. While brain signal processing and its applications have been widely studied for many decades, BCI seldom requires attention to practical and efficient brain signal sensing methods. Noninvasive electroencephalogram (EEG) measurements using wet adhesive Ag/AgCl electrodes are universally employed for BCI, but have limitations in practical acceptability with regard to portability, comfort and aesthetic design. In order to translate the results of laboratory experiments into practical use, EEGs should be recorded easily without requiring scalp preparation and regardless of the presence of hair.

Editorial: Advanced physical methods in brain research

The brain is the most complex organ of the human body. It is located in the head and is the center of the nervous system. Its largest part is the cerebral cortex, which contains tens of billions neurons. On average, each neuron is connected to other neurons through several thousand synapses, which pass electrical (or chemical) signals, to form a very intricate and extremely complex system. The brain is estimated to contain from 1 to 5× 10 synapses. The cerebral cortex is the main characteristic of the human brain (when compared to the brains of other mammals). It consists of a thick layer of neural tissue, folded in a way that maximizes the surface that can fit into the volume available (the skull, whose thick bones provide the necessary physical protection). Each side of the brain is divided into four sections, or “lobes”, called frontal lobe, parietal lobe, temporal lobe, and occipital lobe. Each of them is associated with specific functions. The so-called “executive functions”, such as attention, working memory, planning, organisation, as well as reasoning and abstract thought, are located in the more dorsolateral portions of anterior frontal lobes, while more phylogenetically “ancient” functions like emotions, risk evaluation and impulse control are situated in more medial areas belonging to the limbic system. Continuing with a rough subdivision of work: more posterior parts of the frontal lobes are in charge of voluntary movement; parietal lobes of the relationship between body and space (visuo-spatial functions); temporal lobes of memory and language (the latter lateralised to the dominant emisphere), and occipital lobes of vision. Complex association networks integrate information from the various brain regions to produce percept, thought and action. Brain structures and their functions have always fascinated scientists, philosophers and thinkers. What makes the brain special, from a wider perspective, is that its physical properties give rise to mental states and in this sense they generate what philosophers call the mind. Through more than two thousand years of Western philosophy the mind was thought to be separate from the brain. This mind-body dualism can be traced back to Plato [1] and was formulated in its modern connotation by Rene Descartes five centuries ago [2]. Despite rapid scientific progress, the mind’s connection with brain activity is not thoroughly understood and much about how the brain works remains a mystery, even nowadays. The way individual brain cells and synapses cooperate, giving rise to thought and other superior functions, has challenged biology and medicine, as well as physics and computer science. The approach to human brain damage and disease has been more successful, maybe because they are easier to define than abstract thought. The human brain is susceptible to different types of damage and disease, with different degrees of “visibility” of the pathogenetic pathway. On the one hand, there are obvious dramatic events, such as a blow to the head, a stroke, or poisoning by neurotoxins. On the other hand, there are more insidiously progressive degenerative disorders, such as Parkinson’s disease, or Alzheimer’s disease: conditions characterised by neuronal death and loss of brain volume. Finally, there are psychiatric conditions, such as schizophrenia and depression, that are also associated with brain dysfunction, although the nature of the brain abnormalities in these conditions is more subtle, and less well understood. The techniques used to study the brain can be either invasive or non-invasive. For obvious ethical reasons, the former ones are seldom performed with humans. Non-invasive techniques include neuroimaging or electroencephalography. This Focus Point of The European Physical Journal Plus contains four articles and is devoted to selected open problems in neuroscience, with the emphasis on the advanced brain signal and image processing. During the last years, the development of neuroimaging and signal processing has made the visualization and measurement of pathological brain changes in vivo possible, producing a radical change, not only in the field of scientific research, but also in the

Detecting the Stable, Observable and Controllable States of the Human Brain Dynamics

A new technique is proposed in this paper for real-time monitoring of brain neural activity based on the balloon model. A continuous-discrete extended Kalman filter is used to estimate the nonlinear model states. The stability, controlla- bility and observability of the proposed model are described based on the simulation and measured clinical data analysis. By introducing the controllable and observable states of the hemodynamic signal we have developed a numerical tech- nique to validate and compare the impact of brain signal parameters affecting on BOLD signal variation. This model increases significantly the signal-to-noise-ratio (SNR) and the speed of brain signal processing. A linear-quadratic regulator (LQR) also has been introduced for optimal control of the model.

Alarm clock using sleep analysis

This study is to implement an alarm clock that sounds based on the sleep state of the user. Using known scientific knowledge of the human sleep cycle and its characteristic brain activity, the clock sounds its alarm when the user is in an easily aroused state within a time period set by the user. The user will feel more refreshed than, if they were awakened by a conventional alarm clock (George and Humberto, 2001). There is also a function which helps in recording the hypnogram. The logical processing of brain signals was achieved through National Instruments’ LABVIEW.

Processing of Brain Signals by Using Hemodynamic and Neuroelectromagnetic Modalities

Human neocortical processes involve temporal and spatial scales spanning several orders of magnitude, from the rapidly shifting somatosensory processes characterized by a temporal scale of milliseconds and a spatial scales of few square millimeters to the memory processes, involving time periods of seconds and spatial scale of square centimeters. Information about the brain activity can be obtained by measuring different physical variables arising from the brain processes, such as the increase in consumption of oxygen by the neural tissues or a variation of the electric potential over the scalp surface. All these variables are connected in direct or indirect way to the neural ongoing processes, and each variable has its own spatial and temporal resolution. The different neuroimaging techniques are then confined to the spatiotemporal resolution offered by the monitored variables. For instance, it is known from physiology that the temporal resolution of the hemodynamic deoxyhemoglobin increase/decrease lies in the range of 1-2 seconds, while its spatial resolution is generally observable with the current imaging techniques at few millimeter scale. Today, no neuroimaging method allows a spatial resolution on a millimeter scale and a temporal resolution on a millisecond scale. Nevertheless, the issue of several temporal and spatial domains is critical in the study of the brain functions, since different properties could become observable, depending on the spatiotemporal scales at which the brain processes are measured. It is well knownm that the electroencephalography (EEG) and magnetoencephalography (MEG) are useful techniques for the study of brain dynamics, due to their high temporal resolution. However, it has been said that the spatial resolution of the EEG is rather low, due to the different electrical conductivities of brain, skull, and scalp that markedly blur the EEG potential distributions, making the localization of the underlying cortical generators problematic. In the last ten years, a body of mathematical techniques, known as high-resolution EEG, was developed to estimate precisely the cortical activity from noninvasive EEG measurements. Such techniques include the use of a large number of scalp electrodes, realistic models of the head derived from magnetic resonance images (MRIs), and advanced processing methodologies related to the solution of the so-called “inverse problem,” that is, the estimation of the brain activity (i.e., electromagnetic generators) from the EEG/MEG measurements. The approach implies both the use of thousands of equivalent current dipoles as a source model and the realistic head models, reconstructed from magnetic resonance images, as the volume conductor medium. The use of geometrical constraints on the position of the neural source or sources within the head model generally reduces the solution space (i.e., the set of all possible combinations of the cortical dipoles strengths). An additional constraint is to force the dipoles to explain the recorded data with a minimum or a low amount of energy (minimum-norm solutions). The solution space can be further reduced by using information deriving from hemodynamic measures (i.e., fMRI-BOLD phenomena) recorded during the same task. The rationale of a multimodal approach is that neural activity, modulating neuronal firing and generating EEG/MEG potentials, increases glucose and oxygen demands. This results in an increase in the local hemodynamic response that can be measured by functional magnetic resonance images (fMRIs). Hence, fMRI responses and cortical sources of EEG/MEG data can be spatially related, and the fMRI information can be used as a prior in the solution of the inverse problem. As a result of all these computational approaches, it is possible to estimate the cortical activity with a spatial resolution of few millimeters and with a temporal resolution of milliseconds from noninvasive EEG measurements. In the framework of a COST Action BM0601 NeuroMath, there was organized a workshop in Rome in 2009 on the themes of the processing of neuroelectromagnetic and hemodynamic signals. Selected papers from this conference were subjected to standard peer-review and compiled in this special issue. With this issue we want to illustrate ongoing and emerging research in the development and application of mathematical methods to the recording, analysis, integration, and modeling of neural activity. The selected papers, written by world class scientists, cover diverse issues ranging from computational models to concrete applications of the methods within the neurosciences. We hope that the readership of CIN could appreciate this special issue as we appreciated it during its composition. Laura Astolfi Sara Gonzalez Andino Fabrizio De Vico Fallani Fabio Babiloni

Invasive or Noninvasive: Understanding Brain-Machine Interface Technology [Conversations in BME

With this issue of the magazine, we are adding a new feature, "Conversations in BME," in which distinguished academics and researchers discuss a biomedical issue in depth, highlighting pros and cons of different approaches. Our goal for this feature is to promote discussion as a way to facilitate scientific growth in our community and, in particular, among students. Noninvasive and invasive EEG and electrocorticography (ECoG)-base BMIs were used for brain signal processing.

Probing for local activity‐related modulation of the infrared backscattering of the brain cortex

The possibility to measure the metabolic activity of the brain cortex, with submillimeter spatial and subsecond temporal resolution, would open up enticing scenarios in addressing basic issues on the relation between different structural components of brain signal processing, and in providing an operational pathway to interaction with (dis)functional signal patterns. In the present article, we report the description of a simple system that allows the detection of the minute changes that occur in the optical backscattering of the cortex as a metabolic response to external stimuli. The simplicity of the system is compatible with scalability to an implantable probe. We validate the system on an animal model, and we propose an algorithm to extract meaningful data from the measured signal. We thus show the detection of individual haemodynamic cortical responses to individual stimulation events, and we provide operational considerations on the signal structure.

Development of Si Double-Sided Microelectrode for Platform of Brain Signal Processing System

We have proposed a new implantable neural recording system, which we call the brain signal processing system (BSPS). In this system, LSI chips such as amplifiers, analog-to-digital converters, and multiplexers are integrated on the Si microelectrode array. To analyze the brain functions or to develop medical treatments for brain disorders, a high-density recording of action potentials is strongly required. To realize high-density recording of action potentials, we propose a novel Si double-sided microelectrode that has recording sites on both front and back sides. The back-side recording sites are connected to a recording apparatus by wire bonding through Si via holes. We fabricated the carefully designed Si double-sided microelectrode and evaluated the electrical characteristics of the Si microelectrode. The front- and back-side recording sites had impedance values of 2.5 and 2.7 MΩ at 1 kHz, respectively, which indicated that both recording sites have equivalent characteristics. An in vitro experiment of neuronal action potential recording using the fabricated Si double-sided microelectrode was performed. The CA1 areas of 400-µm-thick hippocampal slices obtained from the brains of guinea pigs were employed, and we successfully recorded neuronal action potentials from the recording sites of both sides.

Estimation of Cortical Connectivity in Humans: Advanced Signal Processing Techniques

In the last ten years many different brain imaging devices have conveyed a lot of information about the brain functioning in different experimental conditions. In every case, the biomedical engineers, together with mathematicians, physicists and physicians are called to elaborate the signals related to the brain activity in order to extract meaningful and robust information to correlate with the external behavior of the subjects. In such attempt, different signal processing tools used in telecommunications and other field of engineering or even social sciences have been adapted and re-used in the neuroscience field. The present book would like to offer a short presentation of several methods for the estimation of the cortical connectivity of the human brain. The methods here presented are relatively simply to implement, robust and can return valuable information about the causality of the activation of the different cortical areas in humans using non invasive electroencephalographic r cordings. The knowledge of such signal processing tools will enrich the arsenal of the computational methods that a engineer or a mathematician could apply in the processing of brain signals. Table of Contents: Introduction / Estimation of the Effective Connectivity from Stationary Data by Structural Equation Modeling / Estimation of the Functional Connectivity from Stationary Data by Multivariate Autoregressive Methods / Estimation of Cortical Activity by the use of Realistic Head Modeling / Application: Estimation of Connectivity from Movement-Related Potentials / Application to High-Resolution EEG Recordings in a Cognitive Task (Stroop Test) / Application to Data Related to the Intention of Limb Movements in Normal Subjects and in a Spinal Cord Injured Patient / The Instantaneous Estimation of the Time-Varying Cortical Connectivity by Adaptive Multivariate Estimators / Time-Varying Connectivity from Event-Related Potentials

Efficient independent component analysis

Independent component analysis (ICA) has been widely used for blind source separation in many fields, such as brain imaging analysis, signal processing and telecommunication. Many statistical techniques based on M-estimates have been proposed for estimating the mixing matrix. Recently, several nonparametric methods have been developed, but in-depth analysis of asymptotic efficiency has not been available. We analyze ICA using semiparametric theories and propose a straightforward estimate based on the efficient score function by using B-spline approximations. The estimate is asymptotically efficient under moderate conditions and exhibits better performance than standard ICA methods in a variety of simulations.

Interrelations of age, sensory functions, and human brain signal processing.

Disturbances of sensory functions may additionally impair cognitive functions in elderly persons. To delineate the impact of visual sensory functions on brain signal processing during normal aging from an electrophysiological perspective, we investigated 289 fit community-dwelling subjects (162 men [56%] and 127 women [44%]; age range, 18-98 years) by means of visual event-related P300 potentials. By taking age, visual acuity, and stimulus-dependent components of visual-evoked potentials (PVEPs) into account, we found age to be the single most important factor for P300 variability (partial F > 10.0, p < .0001 for all P300 parameters; stepwise regression analysis). Furthermore, both the N75 (partial F = 12.415) and P100 latencies (partial F = 4.850) of the PVEPs were independently correlated with the P300 latencies, whereas the P300 amplitudes revealed additional correlations with the P100 amplitudes (partial F = 8.576; p < .0001, for all). Sex, however, did not influence these age-related P300 changes. Aging itself accounts for the largest proportion of variability in human brain signal processing as reflected by P300 potentials. Visual sensory functions, however, also provide an independent, but minor, contribution to P300 variability. Therefore, it seems prudent to take parameters of sensory functions into account especially for clinical P300 applications in the elderly population.

Somatostatin and substance P-like immunoreactivity in the auditory brainstem of the adult rat

The superior olivary complex (SOC) of the adult rat brainstem was studied in detail with regard to its innervation by neural elements showing immunoreactivity for two neuroactive peptides, somatostatin and substance P. Nerve fibres and varicosities showing positive immuno-reactivity for both peptides were particularly dense immediately dorsal and lateral to the lateral superior olivary nucleus (LSO) and dorsal to the superior paraolivary nucleus (SPN). Penetration of this curtain-like innervation into the SPN was limited, and the LSO showed only a very minor innervation by somatostatin-positive structures in its most medial (high frequency) lobe. Dense fibre labelling and varicosities were also apparent for both peptides immediately medial to the ventral and dorsal nuclei of the lateral lemniscus, and in the external cortex and dorsomedial zones of the inferior colliculus (IC). Labelled fibres and endings were also seen in the granule cell regions of anteroventral cochlear nucleus (AVCN) and the most dorsomedial parts of the dorsal cochlear nucleus (DCN). The majority of cells in the medial nucleus of the trapezoid body (MNTB) showed a prominent innervation by nerve terminals that stained positive for somatostatin only whereas the medial superior olivary nucleus (MSO) was devoid of label for both peptides. The ventral nucleus of the trapezoid body (VNTB) showed sparse but significant innervation by both somatostatin and substance P-positive structures. Hence the VNTB was the only defined nucleus of the SOC to show a significant substance P-positive innervation. Neuronal somata immuno-reactive for somatostatin were found in anteroventral and posteroventral cochlear nuclei (AVCN and PVCN) and the A5 and A7 cell groups adjacent to the LSO and the VNLL and DNLL and in all subdivisions of the inferior colliculus (IC). Somata showing only faint immunoreactivity for substance P were found in the VNLL, AVCN and PVCN. These results suggest a potential role for both peptides in auditory signal processing in the adult rat brain.

The relationship between endothelin-1, event-related P300 potentials, and prognosis in cerebral arteriosclerosis.

To search for a potential role of endothelin-1 (ET-1), a potent vasoconstrictor and presumably neurotoxic 21-amino acid peptide, for dysfunction of brain signal processing and cerebrovascular morbidity in nondemented patients with cerebral arteriosclerosis. Cross-sectional study with longitudinal follow-up. University-affiliated teaching hospital. A total of 106 nondemented patients with significant stenosis of either the internal carotid (cAD, cases; n = 63, mean age +/- SD, 62 +/- 7 years) or peripheral arteries (pAD, disease controls; n = 43, 60 +/- 11 years) were investigated before carotid endarterectomy and bypass surgery, respectively. After a mean follow-up of about 19 months, cerebrovascular morbidity of the cAD and pAD patients was evaluated by phone. Brain signal processing functions by event-related visual P300 potentials; cerebrovascular events by a structured telephone interview; the extent of arteriosclerosis by venous ET-1 concentration. Venous ET-1 levels were elevated in both cAD and pAD patient groups, but to the same degree. In these patients, ET-1 concentration was correlated slightly with diastolic blood pressure (r = .334, P = .0326, stepwise regression). Only in cAD patients with ET-1 levels above the 75th percentile were P300 latencies markedly prolonged compared with their lower ET-1 level counterparts. Furthermore, the P300 latencies of the cAD patients, but not of the pAD patients, correlated positively with venous ET-1 concentration and inversely with pack years of smoking (r = .728, P = .0002; stepwise regression). In contrast to base-line P300 abnormalities and classical risk factors (e.g., hypertension), high ET-1 levels predicted an increased cerebrovascular morbidity of cAD, but not of pAD, patients (P = .0044; Mantel Cox test). In nondemented patients with cerebral arteriosclerosis, endothelin-1 is associated with P300 abnormalities reflecting subclinical dysfunction of brain signal processing. In the long-term, high venous ET-1 levels also appear to predict a higher cerebrovascular morbidity of cAD patients even after carotid endarterectomy.

WAVE PROPAGATION, PATTERN FORMATION AND MEMORY EFFECTS IN LARGE ARRAYS OF INTERCONNECTED CHAOTIC CIRCUITS

We investigate complex dynamic phenomena in arrays composed of interacting chaotic circuits. Such arrays can be thought of as a model of nonlinear phenomena in spatially extended (high-dimensional or infinite-dimensional) systems and active media with potential applications in brain function modelling and signal processing. In this paper we consider a particular structure of the network in which there exist locally double diffusive interactions between the cells. Such a double interaction can be considered as a paradigm and means for understanding very complex interactions existing in real systems where neighboring cells can communicate in various ways. We consider two basic cases where separate cells without coupling exhibit two different types of chaotic behavior. Depending on the connection structure, initial conditions imposed in the cells the array exhibits various kinds of spatially ordered chaotic waves. Patterns of behavior depending on the excitation of the array and the connection structure are studied in some detail. We present results of simulations showing cooperative phenomena. Depending on the connection structure between the cells and their initial states the array can show spatially ordered patterns of behavior. Such orderly spatial patterns resulting in arrays of chaotic elements are referred to as self-organization. We observed travelling target waves and chaotic wavefronts behaving like autowaves. This kind of autowave-like behavior in arrays of chaotic elements is to our knowledge described for the first time.

On-line method for enhancement of electroencephalogram signals in presence of electro-oculogram artefacts using nonlinear recursive least squares technique

A record of the human scalp potentials (EEG) is one of the most commonly used techniques for understanding the brain functions. While recording EEG signals, other undesired signals called artefacts get superimposed on the EEG signals. Of these, electro-oculogram generated by eye movements is found to be the most significant and common. Conventional artefact minimisation schemes use linear estimation models. However, the result in sub-optimal schemes as signal processing in the human brain is highly nonlinear. In this paper therefore an on-line technique is suggested for artefact minimisation using nonlinear recursive least squares. Results confirm the efficacy of the method.

Can the dipole tracing method estimate the auditory signal processing in the human brain?

Can the dipole tracing method estimate the auditory signal processing in the human brain?

Desensitization by antidepressants of central norepinephrine receptor systems coupled to adenylate cyclase.

The experimental results discussed in this paper provide evidence that antidepressant-induced attenuation of the NE receptor-coupled adenylate cyclase system in brain and the down-regulation of its beta adrenoceptor subpopulation result in a net deamplification of the NE signal. The desensitization of the NE receptor system requires an unhindered occupancy of the receptor by the agonist NE. Following adrenalectomy, the non-beta population of NE receptors coupled to adenylate cyclase shows an enhanced response to NE without changes in the activity of adenylate cyclase or phosphodiesterase. This supersensitivity to NE can be prevented by corticosterone. The synaptic availability of 5HT is co-required for the down-regulation by DMI-like drugs of the density of beta adrenoceptors. Moreover, beta adrenoceptors from tissue deprived of serotonergic neuronal input display a marked decrease in agonist affinity as determined from competition binding of (-)-isoproterenol for [3H]dihydroalprenolol. Using NE as an agonist, competition binding curves with membrane preparations from cortical tissue lacking 5HT input show low affinity binding of the receptor for NE that cannot be further modified by guanine nucleotides. The reduction in beta adrenoceptor agonist affinity following reduction of the synaptic availability of 5HT is accentuated by chronic administration of DMI or zimelidine. The new experimental data on the biomolecular linkage between serotonergic and noradrenergic neurons, expressed functionally at the level of NE receptors, provide the scientific basis for a "serotonin-norepinephrine link hypothesis" of affective disorders. The pursuit of studies on the molecular mechanisms of the action of steroid hormones on central NE receptor systems and on mechanisms underlying the functional 5HT-NE linkage and its modification by antidepressants should generate a deeper understanding of neuronal signal processing in brain.