Information

What would EEG recordings reveal if their resolution was better?

What would EEG recordings reveal if their resolution was better?


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

1. Temporal resolution

Of course, we would see higher frequencies. But would these be informative? Aren't the frequencies measured today near the border to noise?

2. Spatial resolution

I guess LFP measurement goes in this direction (closer to the sources of the signals being measured). But what would we learn by that? Finding small-scale regions (below Brodman area size) which behave phase-locked/synchronously?

3. fMRI

The same questions could be asked with respect to fMRI. But I guess, as long as temporal resolution is as poor as it is today (and which cannot be much better), we don't have to ask for spatial resolution, or do we?


The temporal resolution of EEG is already considered to be very good. The problem is spatial resolution. Even the loss of very high frequency activity, like individual spikes, is really a problem of spatial rather than temporal resolution: the spiking cells are too far away, so it is only possible to detect them if they are very synchronized (in which case they still show up as lower frequency signals).

Because EEG recordings are so far from the sources they are recording from, and because even structures deep in the brain can contribute to surface potentials,the signals end up as complex averages over a large space. There is an entire field called EEG source modeling that tries to use signal processing methods to improve this spatial resolution.

However, you don't really have to ask this as a hypothetical question, since we already have methods that have better resolution than EEG. Electrocorticograms (ECoG) are the most similar, and most applicable to human studies. ECoG arrays are very similar to EEG, but they are placed directly on the dura, under the skull. They increase spatial resolution because they cut down distances dramatically by not having to record through the skull.

As you mention in your question, local field potentials (LFP) get even closer to where the measured currents are occurring, by recording from within the brain itself. The spatial resolution of LFPs themselves can also be improved by recording from multiple electrodes and computing a current source density. This also helps mitigate the effects of volume conduction.

fMRI is a completely separate issue. Often, EEG and fMRI results can act in conjunction, using EEG to determine time and fMRI to determine space (note that although there are limits to the spatial resolution of fMRI, for a human subject it is dramatically more spatially precise than EEG, especially for any structures not at the very surface of the brain, including any of the sulci of neocortex). fMRI spatial resolution can be improved by using stronger magnets, but eventually there are technological (and expense) limitations, but as you say the temporal resolution is fixed, and even the spatial resolution has some practical limitations.

To generally answer your title question: if the spatial resolution of EEG is improved (practically, this means either source modeling or using ECoG or LFP), one is better able to identify the brain structures that are producing the signals observed in the EEG, and able to be more confident that the signals observed are "real" rather than an artifact of filtering over space. For example, a higher-frequency traveling wave could appear as a lower-frequency signal on an EEG, or a localized, high-intensity event could appear like a generalized, low-intensity event; higher spatial resolution would allow you to distinguish between the two in both cases.


Bin He

Dr. Bin He received his BS in Electrical Engineering from Zhejiang University in 1982, and PhD in Bioelectrical Engineering from Tokyo Institute of Technology, Japan, a Nobel Prize winning campus in 1988, both with the highest honors. He completed a postdoctoral fellowship in Biomedical Engineering at Harvard University - M.I.T. After working as a Research Scientist at M.I.T., he was on the faculty of Electrical Engineering and Bioengineering at the University of Illinois at Chicago, where he was named a University Scholar by the President of the University of Illinois. From January 2004, he has been a Professor of Biomedical Engineering and Director of Biomedical Functional Imaging and Neuroengineering Laboratory at the University of Minnesota. Later he was appointed as a Distinguished McKnight University Professor and Medtronic-Bakken Chair for Engineering in Medicine. He has served as the Director of the Center for Neuroengineering, the NSF IGERT Neuroengineering Training Program, and the NIH Neuroimaging Training Program, and served as the Director of the Institute for Engineering in Medicine from 2012-2017, at the University of Minnesota. Dr. He also served as Director of Undergraduate Studies in Biomedical Engineering from 2004-2006.

Dr. He's research interests include neuroengineering, functional biomedical imaging, cardiovascular engineering, and biomedical instrumentation. He has published over 230 articles in peer-reviewed core international journals including Neuron, Brain, Journal of Neuroscience, Proceedings of the IEEE, NeuroImage, Human Brain Mapping, Nanomedicine, Heart Rhythm, Epilepsia, Applied Physics Letters, American Journal of Physiology, Journal of Neural Engineering, and various IEEE Transactions, including IEEE Transactions on Biomedical Engineering, IEEE Transactions on Medical Imaging, and IEEE Transactions on Neural Systems and Rehabilitation Engineering. He has also delivered over 40 plenary and keynote talks in various international conferences and workshops. Dr. He’s research has been featured by Nature, New York Times, BBC, CNN, NBC, CBS, ABC, Scientific American, Economist, New Scientist, US News, NPR, among others. A video describing his work on mind controlled flying robot has been viewed

900,000 times.

Dr. He has been named the recipient of 2017 IEEE Technical Field Award of Biomedical Engineering, and is a recipient of the Academic Career Achievement Award and the Distinguished Service Award from the IEEE Engineering in Medicine and Biology Society, the Outstanding Research Award from the International Federation of Clinical Neurophysiology, the Established Investigator Award from the American Heart Association, and the CAREER Award from the National Science Foundation, among others. He is an elected Fellow of the International Academy of Medical and Biological Engineering, IEEE, the American Institute for Medical and Biological Engineering, and the Institute of Physics. Dr. He serves as an Associate Editor or Editorial Board Member of multiple international journals in the field of biomedical engineering, and is the Editor-in-Chief of IEEE Transactions on Biomedical Engineering. He has served as the General Chair of the International Annual Conference of IEEE Engineering in Medicine and Biology Society (2009), Chair of the IEEE EMBS Forum on Grand Challenges in Neuroengineering (2010), Co-Chair of Scientific Committee of the World Congress on Medical Physics and Biomedical Engineering (2012), Chair of IEEE Life Sciences Grand Challenges Conference (2012), Chair of IEEE EMBS International Conference on Neural Engineering (2013), Chair of NSF Workshop on Mapping and Engineering the Brain (2013), and Chair of IEEE EMBS BRAIN Grand Challenges Conference (2014). He is a past president of the International Society of Bioelectromagnetism and of the International Society for Functional Source Imaging, and was the 2009-2010 President of IEEE Engineering in Medicine and Biology Society, consisting of 10,000+ members worldwide. He has been a Member of the NIH BRAIN Multi-Council Working Group.

Dr. He has been fortunate to be associated with talented graduate and undergraduate students, and postdoctoral associates, many of whom have taken faculty positions in academia in the US, Canada, Japan, China and Korea, and in major corporations. Many of his PhD students received competitive fellowship awards from various funding agencies or the University, or paper competition awards in international conferences.

Dr. Bin He's major research interests are in the field of neuroengineering and biomedical imaging. His research programs are funded by NIH (NIBIB, NCCIH, NINDS, NHLBI, NEI, OD), NSF, and ONR, among other sponsors. The active research programs in Dr. He's lab are as follows.

Multimodal Functional Neuroimaging: Dr. He and his students have developed a unified theory for multimodal neuroimaging integrating the BOLD functional MRI and electrophysiological imaging. Hemodynamic neuroimaging, such as BOLD functional MRI, has high spatial resolution at mm scale but very slow in time. Electrophysiological neuroimaging has high temporal resolution at ms scale but limited spatial resolution. It has been a major frontier in the functional neuroimaging research to attempt to greatly enhance spatio-temporal resolution by integrating functional MRI with electrophysiological imaging. Dr. He and colleagues have developed a rigorous theory on neurovascular coupling, which provides a principled way of integrating BOLD functional MRI signals with electrophysiological signals for event-related paradigms. The theory has been tested in human visual system and revealed a dramatic improvement in performance in imaging human visual information pathways. Dr. He and his students have further developed new algorithms to integrate fMRI and EEG/MEG signals for oscillatory brain activity. Currently both theoretical and experimental studies are actively pursued in Dr. He's lab to further develop the high resolution spatio-temporal functional neuroimaging modality, and to study the sensory, motor and cognitive functions of the brain using EEG/fMRI.

Functional Neuroimaging of Epilepsy: Dr. He and colleagues have made significant contributions to high-resolution electrophysiological neuroimaging aiding neurosurgical planning in epilepsy patients. Due to the limited spatial resolution of scalp EEG, it is widely practiced in clinical settings that invasive intracranial recordings are obtained, by placing electrode sensors directly on the surface of or within the brain, to aid neurosurgical planning in patients with intractable epilepsy. Dr. He and his students have developed an innovative epilepsy source imaging methodology, in which causal interactions among sources are identified and imaged from noninvasive EEG recordings. Such imaging provides high spatial resolution in imaging distributed brain sources within the 3-dimensional brain volume and reveals neural interactions and connectivity embedded in the brain networks. Dr. He and colleagues have further developed an ICA based seizure imaging methodology and conducted a rigorous validation study in a group of epilepsy patients to image epileptogenic brain, and demonstrated high consistence between the imaged seizure sources and the epileptogenic zones determined by well established clinical procedures in the same patients. Active research is currently being pursued to further establish electrophysiological neuroimaging as a noninvasive tool aiding surgical planning in epilepsy patients. Other than EEG/MEG neuroimaging, Dr. He and his colleagues are also pursuing functional MRI mapping of epileptogenic zone from fMRI. This line of work is carried in collaboration with Mayo Clinic and University of Minnesota Medical Center.

Electrical Properties Imaging: An important research program in Dr. He’s lab is the development and investigation of novel approaches for noninvasive imaging of electrical properties of biological tissues, including bioimpedance imaging. Dr. He and colleagues have proposed and developed a new approach called magnetoacoustic tomography with magnetic induction (MAT-MI) by integrating ultrasound and biomagnetism, in order to obtain high resolution image of electrical impedance of biological tissue. In the MAT-MI approach, the object is placed in a static magnetic field and a pulsed magnetic field. The pulsed magnetic field induces eddy current in the object. Consequently, the object emits ultrasonic waves through the Lorentz force produced by the combination of the eddy current and the static magnetic field. The acoustic waves are then collected by the detectors located around the object for image reconstruction. MAT-MI takes the advantage of excellent contrast of electrical impedance and the high spatial resolution of ultrasound. The recent work demonstrates that the MAT-MI approach can achieve mm spatial resolution in imaging electrical impedance, which represents a significant advancement in comparison with the spatial resolution of conventional electrical impedance imaging approach, and may play a critically important role in early detection of breast cancer. Along the similar line, Dr. He and his colleagues have been developing a MR based electrical properties tomography (EPT) approach using B1 mapping of MR technology to reconstruct subject specific electrical properties distributions. EPT not only has great potential for clinical applications in cancer detection and diagnosis but also promises to provide subject specific SAR mapping, helping managing safety concerns in high or ultrahigh field MRI systems. The EPT project is carried in collaboration with CMRR investigators.

Cardiac Electrical Tomography: Dr. He and colleagues have pioneered the development of electrophysiological cardiac tomography in assessing dynamic cardiac functions. The electrocardiographic inverse imaging problem has historically been solved using point dipole sources, epicardial potentials, or heart surface activation patterns. Dr. He and colleagues have developed cardiac electrical tomography techniques in his lab to image electrical functional information throughout the 3-dimensional volume of the heart from noninvasive electrocardiographic measurements. In collaboration with collaborators at the University of Alabama at Birmingham and the University of Minnesota Medical School, Dr. He and colleagues have validated this cardiac electrical tomography technology in animal models including rabbits, dogs, and swine where simultaneous intracardiac or intracavity potentials are measured together with body surface potential mapping. Currently pilot clinical studies are being carried out to test the efficacy and clinical applications of the cardiac electrical tomography techniques developed in Dr. He’s lab. The establishment of such imaging techniques promises to greatly impact the management of cardiac arrhythmia, a leading cause of public health problem in US and developed countries.

Brain-Computer Interface: Dr. He has made pioneering contributions to noninvasive brain-computer interface (BCI) research. This work is aimed at developing novel techniques for effectively decoding the intention of human subjects and controlling external device, which may ultimately benefit patients suffering from neurological disorders such as spinal cord injuries or stroke. Dr. He and his colleagues have developed a time-frequency-spatial approach to extract the extremely weak signals accompanying the “thought” of a human subject using an array of electrode sensors placed over the scalp. This method takes the “signatures” of each individual subject and uses them for optimal decoding of the intention of the human subject. Dr. He has proposed the concept of electrophysiological neuroimaging based BCI – an idea to estimate “virtual” intracranial signals from the noninvasive EEG recordings for substantially improving the performance of noninvasive EEG based BCI. Dr. He and his colleagues have been aggressively investigating the mechanisms associated with motor imagery based BCI by using advanced neuroimaging techniques to delineate the brain sources accompanying motor imagery. Recently, Dr. He and his students have developed a 3-dimensional continuous brain-computer interface system to allow human subjects to control the flight of a flying robot from noninvasive brain waves. Click the video to see the mind controlled flying robot.

Neuromodulation: Dr. He and his colleagues are actively pursuing noninvasive neuromodulation modalities including transcranial focused ultrasound (tFUS), transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), and transcranial alternate current stimulation (tACS). Current interest ranges from basic studies of mechanisms of tFUS, TMS and tDCS/tACS by means of dynamic brain mapping, to clinical investigations in patients with stroke and schizophrenia. Other than treatment purposes, Dr. He and his students are pursuing perturbation based imaging in which neuromodulation is used to perturb the central nervous systems and responses measured using various neuroimaging methods for better understanding of brain circuits, networks and functions.


Role of EEG as Biomarker in the Early Detection and Classification of Dementia

The early detection and classification of dementia are important clinical support tasks for medical practitioners in customizing patient treatment programs to better manage the development and progression of these diseases. Efforts are being made to diagnose these neurodegenerative disorders in the early stages. Indeed, early diagnosis helps patients to obtain the maximum treatment benefit before significant mental decline occurs. The use of electroencephalogram as a tool for the detection of changes in brain activities and clinical diagnosis is becoming increasingly popular for its capabilities in quantifying changes in brain degeneration in dementia. This paper reviews the role of electroencephalogram as a biomarker based on signal processing to detect dementia in early stages and classify its severity. The review starts with a discussion of dementia types and cognitive spectrum followed by the presentation of the effective preprocessing denoising to eliminate possible artifacts. It continues with a description of feature extraction by using linear and nonlinear techniques, and it ends with a brief explanation of vast variety of separation techniques to classify EEG signals. This paper also provides an idea from the most popular studies that may help in diagnosing dementia in early stages and classifying through electroencephalogram signal processing and analysis.

1. Introduction

Dementia refers to a group of disorders caused by the gradual dysfunction and death of brain cells. This disorder can be described clinically as a syndrome that causes a decline in cognitive domain (i.e., attention, memory, executive function, visual-spatial ability, and language) [1]. Predicting dementia in the early stages would be essential for improving treatment management before brain damage occurs.

The early diagnosis of dementia will help dementia patients start an early treatment based on the symptoms. In the past years, significant advances have been made to reveal the early stages of dementia through biomarkers. These improvements include biochemical, genetic, neuroimaging, and neurophysiological biomarkers [2, 3]. Therefore, developing and integrating these biomarkers to identify dementia in early stages are important to derive an optimal diagnostic index.

In parallel, over the last two decades, significant growth was noted in the research interest on EEG, as the full investigation of neurodynamic time-sensitive biomarker that helps in detecting cortical abnormalities associated with cognitive decline and dementia [4–7]. An EEG marker would be a noninvasive method that may have the sensitivity to detect dementia early and even classify the degree of its severity at a lower cost for mass screening. EEG is also widely available and faster to use than other imaging devices [8, 9].

This review has focused on using EEG as an investigating tool and physiological biomarker to identify dementia in early stages and classify the degree of its severity by signal processing and analysis. The review aims to reveal subtle changes that might define indicators for the early detection of dementia that will help medical doctors and clinicians in planning and providing a more reliable prediction of the course of the disease in addition to the optimal therapeutic program to provide dementia patients additional years of a higher quality of life.

2. Dementia and Medical Diagnosis

Dementia occurs when the brain has been affected by a specific disease or condition that causes cognitive impairment [10]. The diagnosis of dementia is usually based on several criteria, such as the medical history of patients with clinical, neurological, and psychological examination, laboratory studies, and neuroimaging [3].

2.1. Types of Dementia and Cognitive Spectrum

Dementia is associated with neurodegenerative disorder diversity, as well as neuronal dysfunction and death. Dementia has different types based on its cause these types include Alzheimer’s disease (AD), vascular dementia (VaD), Lewy body, frontotemporal dementia (FTD), and Parkinson’s disease, among others [2, 11].

AD and VaD are considered the two most common types of dementia in the world, and thus the present review deals with the effect of AD and VaD on the brain [12]. AD is the most prevalent in the Western world, whereas VaD is the most prevalent in Asia [13].

Half of people aged 85 years or older have AD, and this number will roughly double every 20 years due to the aging population [14, 15]. Several neuropathological changes act together to develop AD. These changes include loss of neuronal cell and development of neurofibrillary tangles and amyloid plaques in the hippocampus, entorhinal cortex, neocortex, and other regions of the brain. These changes can also occur in a nondemented individual, and they are associated with AD development even before typical cognitive symptoms are evident [16, 17]. The reduction in cholinergic tone caused by neural damage results in an increase in cognitive difficulties [18].

VaD is another type of dementia. Between 1% and 4% of people aged 65 years suffer from VaD, and the prevalence for older people doubles every 5 to 10 years [19, 20]. VaD is the loss of cognitive function caused by ischemic, ischemic-hypoxic, or hemorrhagic brain lesions as a result of cerebrovascular disease and cardiovascular pathologic changes, such as ischemic heart disease and stroke [21–23].

Cognitive impairment introduces individuals to the dementia spectrum that is illustrated in Figure 1. The dementia spectrum can be viewed as a sequence in the cognitive domain that starts from mild cognitive impairment (MCI) and ends with severe dementia, and the period beyond dementia in which the brain is at risk is called cognitive impairment no dementia (CIND) [24].


MCI refers to the decline in cognitive function that is greater than expected with respect to the age and education level of an individual, but the reduced cognitive function does not interfere with daily activities. Clinically, MCI is the transitional stage between early normal cognition and late severe dementia and is considered heterogeneous because some MCI patients develop dementia, whereas others stay as MCI patients for many years. However, patients who were diagnosed with MCI have a high risk to develop dementia, that is, threefold that of people without a cognitive dysfunction. The most commonly observed symptoms of MCI are limited to memory, whereas daily activities of patients remain the same [25].

As dementia diagnosis is not easily performed due to the heterogeneity of the symptoms within the cognitive impairment spectrum, it may be advisable to integrate the neuropsychological testing with biomarkers. The latest diagnosis criteria for AD and MCI support this idea as they highlight the importance that several biomarkers (structural MRI, FDG-PET, and biochemical analyses of the cerebrospinal fluid) have to confirm that a pathological process of AD is, indeed, the cause of the cognitive symptoms [26–29]. The diagnosis criteria usually focus of assessing diverse dementia signs, particularly memory disturbance. The most common diagnosis criteria are developed and characterized by the National Institute of Neurological and Communication Disorder and Stroke-Alzheimer’s Disease and Related Disorder Association (NINDS-ADRDA) for AD [26–30] and the National Institute of Neurological Disorders and Stroke and Association Internationale pour la Recherché et l’Enseignement en Neurosciences (NINCDS-AIREN) for VaD [31] and Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM-IV) criteria [32]. The severity of cognitive symptoms could be assessed using Clinical Dementia Rating (CDR) scale [33] and Geriatric Depression Scale (GDS) [34] and Hachinski Ischemic Scale (HIS) [35], whereas the functional outcome can be assessed by instrumental and basic activity of daily living (IDAL) and (BDAL) [36]. The most usable tests to evaluate the early dementia stages even severity of dementia in clinical practice are the Mini-Mental State Examination (MMSE) [37], Montreal Cognitive Assessment (MoCA) [38], and Addenbrooke’s Cognitive Examination Revised (ACE-R) [39]. Several validate clinical neuropsychological assessments are used to assess cognitive domain including (but not limited to) Trail Making Test (TMT) [40] and Clock Drawing Test (CDT) [41] for attention and executive function, Rey Osterrieth Figure Copy [42] for construction praxis test, and Phonological and Semantic fluency Token test for language test [43].

2.2. Biomarkers for Detecting Dementia

An objective measure, which is related to molecules that are concentrated in the brain or biological fluids or to other anatomical or physiological variables, that help in diagnosing and assessing the progression of the disease or the response to therapies is called a “biomarker.” A biomarker can be used to view the pathogenesis of dementia and helps predict or evaluate the disease risk to identify a clinical diagnosis or therapeutic intervention monitoring that may alter or stop the disease [44, 45]. Ideally, the biomarker should detect the neuropathological processes even before a clinical diagnosis and should help in identifying people who are at risk of developing dementia. The biomarkers for the early detection of dementia may include numerous studies in multiple fields and may be divided into four main categories, namely, biochemical, genetic, neuroimaging, and neurophysiology [2, 3, 11, 46].

2.2.1. Biochemical Marker

Two main types of biochemical markers were identified to reflect the pathological events, particularly detection of dementia, cerebrospinal fluid (CSF), and serum [2, 47].

Several studies have addressed the development in amyloid β (Aβ), total tau (T-tau), and hyperphosphorylation tau (P-tau) protein analysis in CSF and plasma as biomarkers for AD. Although CSF biomarkers are specific of AD, Paraskevas et al. [48] has investigate their potential contribution for the differential diagnosis between AD, MCI, and VaD. For instance, Aβ42 and T-tau in CSF are useful in differentiating MCI and other dementia stages within the dementia spectrum, whereas the CSF measurement of P-tau and Aβ42 can assist in diagnosing VaD or FTD [49]. However, both CSF and serum are used as markers to identify dementia, but the sensitivity and the specificity of these tests are limited [11].

2.2.2. Genetics Biomarkers

Gene expression profile is considered a promising approach for the early detection of dementia. Several studies have been conducted through the genetic analysis of related disorders, such as AD, to evaluate the genetic risk factor that may lead to dementia. Moreover, blood-based gene expression profiling has been described as capable of diagnosing brain disorders by several independent groups. Numerous advantages are offered by the expression profiling of whole blood RNA in deciphering aberrant patterns of gene regulation in neurogeneration. Therefore, the genetic biomarker provides an indication to develop dementia but also needs other biomarkers, such as neuroimaging and chemical biomarkers [2, 3, 11]. The

allele of the apolipoprotein E gene is the major lipid carrier of protein to the brain, and its inheritance is associated with the onset of AD and VaD. Accordingly, age and the inheritance of the allele have been used as a common risk factor and/or pathogenesis for both AD and VaD [45, 50].

2.2.3. Neuroimaging Biomarkers

Neuroimaging has been available for a few decades. This technique can be classified into structural and functional based on the principal information that it provides. Both magnetic resonance imaging (MRI) and computed tomography (CT) are structural imaging techniques they help clarify the brain diagnosis by detecting the affected area and the type of atrophy or vascular damage. The role of CT is to distinguish two structures and separate them from each other, as CT has good spatial resolution. By contrast, MRI distinguishes the differences between two arbitrarily similar but not identical tissues. MRI provides a good contrast resolution. Positron emission tomography and single photon emission computed tomography are considered functional imaging techniques that can measure brain metabolism parameters, such as regional cerebral blood flow and regional cerebral glucose metabolism. These parameters provide good indication for AD and VaD before morphological changes occur. Moreover, functional MRI is used to measure the brain function over time based on blood oxygen level at rest. It indirectly reflects neuronal activity and identifies the brain activities that are associated with cognitive tasks. Functional imaging techniques are suitable in early dementia detection and diagnosis [2, 3, 11]. These techniques have high spatial resolution for anatomical details but limited temporal resolution. Thus, these neuroimaging techniques are incapable of differentiating the stages within the brain distribution network in series or in parallel activation [51]. Additionally, CT and MRI may be affected by fluid imbibition after brain injury in some cases, thus becoming incapable of detecting the best risk changes or becoming inadequately sensitive to detect dementia in its early stages [52].

2.2.4. Neurophysiological Biomarkers

Neural changes associated with dementia can also be detected with clinical biomarkers, such as EEG, quantitative electroencephalography, event related potential, transcranial magnetic stimulation, and Vagus nerve stimulation [2, 18]. EEG is a neurosignal that tracks information processing with milliseconds precision. It has been subjected to interpretation by clinician visual inspection that results in acceptable and successful diagnosis results. However, EEGs are characterized by spatial resolution that is lower than that of other neuroimaging techniques, although these techniques do not provide functional information about the brain in addition to their limitation in temporal resolution EEG provides high temporal resolution and it is thus crucial for studying brain activity [53, 54]. Thus, the interpretation of the degree of EEG abnormality and severity of dementia are the benefits of signal processing and analysis of EEG. EEG signal analysis provides a relatively precise localization of electrical activity sources by tracking the hierarchical connectivity of neurons in the recording place. EEG may provide useful indication of the patterns of brain activity if it is integrated with other biomarkers, such as structural and functional neuroimaging [51].

With the dramatic progress in EEG devices, sensors, and electrodes, this review has been focused solely on the function of EEG as a subtle and suitable biomarker in explicitly identifying the neuronal dynamics and cognitive manifestation in most dementia cases, such as AD and VaD, through techniques of EEG signal analysis and processing.

3. Function of EEG in the Early Detection and Classification of Dementia

As a neurophysiological biomarker, EEG can characterize different physiological and pathological conditions, such as dementia effects on cortical function distribution. EEG could be used not only as a clinical diagnosis tool, but also as a tool for predicting the stages of dementia [7]. Numerous studies have been conducted to deal with EEG changes associated with dementia and to identify the degree of severity of dementia, and some studies support the possibility for EEG to detect dementia in early stages [55–59]. For instance, Henderson et al. identified dementia presence early through EEG with high sensitivity and specificity [55, 60] they showed the possibility of using EEG as a marker for AD [61]. EEG may play an important role in detecting and classifying dementia because of its significant influence on dementia abnormalities in terms of rhythm activity. EEG is useful for clinical evaluation because of its ease of use, noninvasiveness, and capability to differentiate types and severity of dementia at a cost lower than that of other neuroimaging techniques [8, 9].

3.1. EEG Signal and Mental States

To deal with EEG signals and to extract useful information and features that help in early dementia diagnosis, an EEG signal should be illustrated in terms of its rhythmic activity [9]. A clinical EEG wave forms an amplitude that is typically between 10 and 100 μv and at a frequency range of 1 Hz to 100 Hz. EEG can be classified into the following five rhythms according to their frequency bands as shown in Figure 2. (1) Alpha (α) wave: this rhythmic wave appears in healthy adults while they are awake, relaxed, and their eyes are closed. It occurs at a frequency range of 8 Hz to 13 Hz with a normal voltage range of approximately 20 μv to 200 μv [62]. α waveform is diminished by opening the eye, sudden stimulus and attention, and a phenomenon known as alpha blockage or desynchronization. As α wave distribution and outcome are based on etiology, the EEG patterns which predominate in the α frequency band in case of unconscious or comatose state are defined as alpha coma [63]. α rhythm is composed of subunits including alpha1, alpha2, and alpha3, whose spectral band power gives an indication on dementia severity [64, 65]. α waveform is mostly observed in the posterior region of the head [66]. (2) Beta (β) wave: the frequency of β waves ranges from 13 Hz to 30 Hz, which is higher than that of the α waveform, but their amplitudes are lower and range from 5 μv to 10 μv [62]. β waves appear with extra excitation of the central nervous system, increase with attention and vigilance, and replace α wave during cognitive impairment. β waves are observed in the parietal and frontal region of the scalp [66]. (3) Theta (θ) wave: the frequency range of θ wave is 4 Hz to 7 Hz. This waveform is prominent during sleep, arousal in older children and adults, emotional stress, and idling. θ wave is recorded across the temporal and parietal region of the scalp with an amplitude range of 5 μv to 10 μv [62]. Two types of θ are found among adults based on their activity the first type shows a widespread distribution across the scalp and is associated with decreased alertness, drowsiness, cognitive impairment, and dementia, whereas the second type is called frontal midline theta because it is distributed within the frontal midline and is generated by the anterior cingulated cortex, which is the largest region with a positive correlation between the theta current density and glucose metabolism. This wave has been linked to activities such as focusing, attention, mental effort, and stimulation processing [66]. (4) Delta (δ) wave: the lowest frequency of δ wave is less than 3.5 Hz, and its amplitude ranges from 20 μv to 200 μv. δ wave occurs during deep sleep, in infancy, and with serious organic brain diseases. This waveform can be recorded frontally in adults and posteriorly in children [62]. (5) Gamma (γ) wave: the frequency of γ wave ranges from 30 Hz to 100 Hz [62]. This waveform is recorded in the somatosensory cortex in the case of cross model sensory processing, during short-term memory to recognize objects, sounds, tactile sensation, and in pathological case because of cognitive decline, particularly when it is related to θ band [66].


(a)
(b)
(c)
(d)
(e)
(f)
(a)
(b)
(c)
(d)
(e)
(f) EEG frequency waveform. (a) One second of EEG signal. (b) Delta wave. (c) Theta wave. (d) Alpha wave. (e) Beta wave. (f) Gamma wave.

Until late adulthood, the activities of δ and θ waves diminish with age, whereas those of α and β waves increase linearly [67]. The current density of δ and glucose metabolism possess an inverse relationship in the case of cerebrovascular diseases, such as stroke, and may be found within the subgenual prefrontal cortex as an outcome of dementia cognitive impairment [68].

3.2. EEG Finding in Dementia

EEG has been used as a benchmark for the detection and diagnosis of dementia for two decades. Numerous studies have supported the capability of EEG recording to detect AD and VaD early [59, 69]. Other studies have used EEG as a tool for differentiating AD from other types of dementia, particularly in the differential diagnosis of AD and VaD [70, 71]. EEG can diagnose the two most common types of dementia (i.e., AD and VaD) because both of these types are cortical, and EEG reflects hidden brain abnormalities [72, 73].

The first EEG clinical observation was illustrated by Berger in the beginning of the last century [74, 75]. The interpretation of the conventional visual characteristics related to AD can be summarized by slowing the EEG dominant posterior rhythm frequency, increasing the diffused slow frequency, and reducing both alpha and beta activities, whereas the occipital alpha activity is preserved and theta power is increased in the case of VaD. The delta power is increased in both AD and VaD patients [4, 76]. The computerized EEG signal analysis provides quantitative data, including reduced mean frequency, increased delta and theta power along with decreased alpha and beta power, reduced coherence in the cortical area, and reduced EEG complexity in dementia patients [4]. Numerous studies by Moretti et al. investigate subrhythms within alpha, where the power ratio of alpha3/alpha2 is used as an early marker for prognosis of MCI and the increase in this ratio is correlated with hippocampal atrophy in both MCI and AD patients, whereas theta/alpha1 ratio could be as a reliable index for cerebrovascular damage [64, 77–79]. However, EEG may exhibit normal frequency and may appear similar to normal aged control subjects during the earliest stages of dementia [4]. Nonetheless, EEG signal analysis may contribute to the deeper understanding of dementia because such computerized analysis provides quantitative data instead of mere visual inspection.

3.3. EEG Signal Processing

The recorded EEG needs successive stages of signal processing to extract meaningful markers from the EEG signal of dementia patients, and these markers reflect brain pathological changes. The main stages of EEG signal processing are denoising, feature extraction, and classification. Figure 3 illustrates the stages of EEG signal processing.


3.3.1. EEG Signal Acquisition Stage

EEG is a medical device that reflects the electrical activity of the neurons of the brain and records from the scalp with metal electrode and conductive media [80].

Figure 4 shows the general EEG machine schematic diagram it consists of electrodes, amplifiers, A/D converter, recorder, storage, and display devices.


For dementia patients, several procedures have been proposed to record the EEG signal for instance, the gold plate cup electrodes shown in Figure 5 have been used to record EEGs. The skin should be swabbed with alcohol and gel or paste should be applied before placing the electrode on the scalp to reduce the movement of the device and improve the electrode conductivity the EEG electrode-scalp contact impedance should be below five kilo-ohms to record good quality signal [81].


Referential montage is the most popular montage used for EEG recording for dementia that is employed to record the voltage difference between the active electrode on the scalp and the reference electrode on the earlobe, for example, as shown in Figure 6 [82, 83].


For fruitful clinical application, the EEG of dementia patients has been recorded in a specialized clinical unit state with the 10–20 system of the international federation, which is adopted by the American EEG Society, while resting with eyes comfortably closed, as shown in Figure 7. Hamadicharef et al. [84] used 19-recording electrodes plus ground and system reference for EEG recording for dementia patients these electrodes were located according to 10–20 electrode system as follows: Fp1, Fp2, F7, F3, Fz, F4, F8, A1, T3, C3, Cz, C4, T4, A2, T5, P3, Pz, P4, T6, O1, and O2 [85].


(a)
(b)
(a)
(b) The 10–20 EEG electrodes placement system. (a) and (b) Three-dimensional side view and top view, respectively [85].

An example of the most popular EEG device contains a low pass, high pass, and notch filters. Typical frequency values for low pass filter (LPF) (i.e., 3 dB) are 0.16, 0.3, 1.6, and 5.3 Hz, and the upper cutoff frequency can be, for example, 15, 30, 70, or 300 Hz. Typically, the frequency for EEG recording for dementia range is from 0.3 Hz to 70 Hz, and the notch filter is 50 Hz or 60 Hz [86]. The sampling frequency can be 128 Hz, 173 Hz, or even higher such as 256 Hz and it is selected based on the application with a 12 bit or 16 bit A/D converter digitalizing the signal to be more accurate. Finally, the EEG signal will be printed on papers, displayed on the computer screen, and stored for further examination in the next stage [81].

3.3.2. Denoising Stage

The reliability of the recorded EEG signal is heavily affected by its noise factors. Most artifacts overlap with the frequencies of EEG signals. The artifacts that contaminated the EEG signal are divided into physiological (e.g., muscle activity, pulse, and eye blinking) [87–90] and nonphysiological artifacts (e.g., power line interference noise and sweat) [90, 91] and/or neuronal activity (e.g., background). The noise has a direct effect on EEG signal properties, and thus different signal processing techniques have been applied to overcome this problem and to extract relevant information from the recorded EEG signal. In order to focus on the role of EEG in the diagnosis of dementia, the mathematical details have been simplified in the text. This section discusses the most popular and effective methods used for EEG denoising.

Independent component analysis (ICA) is a blind source separation higher order statistical method used to split a set of recorded EEG signal (i.e., mixed signals) into its sources without previous information about the nature of the signal. Langlois et al. and McKeown et al. used ICA to observe the EEG signal mixture that reflects multiple cognitive activities or artifacts, particularly the ocular artifact [92–94].

Wavelet transform (WT) is an effective denoising procedure that was introduced to process nonstationary signals, such as EEG. Zikov et al., Krishnaveni et al., and other researchers used WT to remove ocular artifact [95–98]. The continuous wavelet transform (CWT) can be used as a set of decomposition functions called mother wavelet the most popular mother wavelets used in biomedical signal denoising are Daubechies, coiflets, and dyme, as shown in Figure 8.


WT is considered a method for multiresolution analysis that provides varying resolutions at different time and frequency [99], as shown in Figure 9.


Nazareth et al. and other researchers applied an effective new approach by combining ICA and WT resulting in an ICA-WT hybrid technique, as shown in Figure 10. As an example, ICA-reconstructed data were cascaded as an input to WT decomposition. This merge assists ICA in distinguishing the signal and noise even if both nearly have the same or higher amplitude and removes overlapping noise signal. Furthermore, the WT can decompose EEG signals into different subbands based on the decomposition levels [100–105]. The ICA-WT technique has illustrated successful results in removing the electrooculography and muscle activity artifacts [105]. Accordingly, this technique is useful in revealing hidden EEG characteristics by the next stage. Thus, the signal is ready for the next stage (i.e., feature extraction stage).


3.3.3. Dementia Feature Extraction and Selection

The denoised EEG signal from the previous stage undergoes feature extraction to detect dementia and develop a useful diagnostic index using EEG. This stage aims to extract the useful information from the EEG of dementia patients by linear and nonlinear techniques.

Linear techniques have been used to extract meaningful features from the EEG of dementia patients that are useful as early dementia indices. Jeong used linear techniques based on coherence and spectral calculations that were used to find EEG abnormalities [4]. A slowdown in EEG signals in dementia is illustrated by the shifting of power to the lower frequency and the decrease in interaction among the cortical area (i.e., increase in delta and theta power along with decrease of alpha power) [106]. Spectral analysis has intensively been used to gain insight into dementia, for instance, Escudero et al. [107] have analyzed the magnetoencephalogram (MEG) signals to quantify their abnormalities in the spectra of dementia patients with two spectral features (i.e., the median frequency (MF) and the spectral entropy (SpecEn)) based on their usefulness in distinguishing the brain activity of dementia patients from the normal-age match subjects. Both spectral features provide information about the relative power of low and high frequencies that reflect the local synchronization of the neural assemblies [108]. Thereafter, the electrical brain activity for dementia patients is characterized by the slowing of brain frequency, and this property can be performed using MF and SpecEn [107].

AD and VaD patients share spectral analysis properties, such as the slowing of alpha power and increase in delta power, but theta power is higher in VaD patients than in AD patients [86]. However, EEG frequencies may look normal in the early stages of AD [109]. Generally, the severity of cognitive impairment and the degree of EEG abnormalities are correlated [4].

EEG coherence is used to evaluate the cortical connection functionality and quantify cortico-cortico or cortico-subcortical connection. Moreover, the coherence function can be used to quantify the linear correlation and detect the linear synchronization between two channels however, this function does not distinguish the directionality of the coupling [110, 111]. A decrease in coherence is interpreted as a reduction in linear function connection and function uncoupling in the cortical area. By contrast, an increase in coherence is interpreted as augmented linear function connection and function coupling in the cortical area [51].

Nonlinear dynamic techniques have been used intensively to analyze the EEG signal, particularly the EEGs of dementia patients, for decades. Researchers have used EEG to investigate the complex dynamic information that is reflected from the brain cortex and recorded by EEG devices [112, 113]. The hypothesis that the brain is stochastic may be rejected based on the capability of the brain to perform sophisticated cognitive tasks thanks to its complicated structure. Moreover, brain neurons are controlled by nonlinear phenomena, such as threshold and saturation processes, such that brain behavior can be classified as nonlinear. The nonlinear dynamic analysis may be considered a complementary approach in detecting mental diseases, because it provides additional information to that of traditional linear methods [114, 115]. Moreover, numerous methods have been introduced to study time series EEG data from human brain activity to understand and detect EEG abnormalities.

The first nonlinear methods that were used to analyze EEG are the correlation dimension (D2) and the first Lyapunov exponents (L1). D2 was applied by Grassberger and Procaccia in 1983 to quantify the number of independent variables that are necessary to describe the dynamic system. It was used to provide the statistical characteristic of the system. By contrast, L1 was applied by Wolf in 1985 as a dynamic measure to gauge the flexibility of the system [116, 117]. Early detection of dementia can be predicted using fractal dimension (FD), zero-crossing interval (ZCI), entropy, such as sample entropy (SampEn) and Kolmogorov entropy, central tendency measure, and Hojorth-Index. Hamadicharef et al. presented the performance results of these methods based on sensitivity, specificity, accuracy, area under the ROC curve, and standard error and found that FD and ZCI are the best methods [118].

Henderson et al. successfully applied FD as a powerful tool for transient detection in terms of waveform that is used to measure the signal structure details in biology and medicine. The derivation of FD of the autocorrelation function can be found in [119]. Moreover, they used ZCI to analyze EEG [55].

Lempel-Ziv-Welch (LZW) is a metric that has been applied to evaluate the signal complexity by measuring the number of distinct substring and their rate of recurrence along the time series Ferenets et al. introduced an algorithm to compute the LZW [120].

Several methods have dealt with the complexity or irregularity in the ability of the system to create information by entropy methods, such as Tsallis entropy (TsEn), approximation entropy, SampEn, and multiscale entropy (MSE) [55, 61, 115, 121–124].

To sum up, linear spectral methods have been used traditionally in the field. Their interpretation may be more straightforward for clinicians, as they are closely related to the power associated with different brain rhythms (alpha, beta, delta, and theta), whereas nonlinear techniques may provide complementary information. Nonlinear methods are motivated by the nonlinear behavior of the neurons in the brain. Both approaches have been used to inspect the EEG activity in dementia, but most studies have focused on only one of those families of methods and there are few comprehensive comparative studies [108]. Despite potentially promising findings, the sizes of the analyzed datasets limit the results. These features are applied to the next stage to estimate the degree of the severity of dementia.

3.3.4. Dementia Classification Techniques

The classification staging is necessary to predict the qualitative properties of the mental state of dementia patients. In this stage, the feature vectors extracted from the previous stage were classified into three categories, namely, CIND, MCI, and dementia. Feature vectors must be analyzed further before being applied to the classifier to avoid overloading the classifier and reduce the computational time, increasing the accuracy of classification. These feature vectors can be processed using dimensionality reduction techniques as shown in Figure 11. Numerous methods can be used including principal component analysis (PCA) and ICA. These methods are well-established methods for dimensionality reduction. PCA is a widely used method to avoid the redundancy because of high-dimensional data [125–127]. The dimensionality-reduced features were used as an input to the classifiers to improve the accuracy of the classification of the severity of dementia by EEG signal analysis.


In EEG applications, highly accurate classification is strongly related to the quality of extracted features, the dimensionality reduction, and the classifiers. Linear discriminant analysis (LDA) and support vector machine (SVM) classifiers are the most popular methods used to classify brain disorders, such as dementia and epilepsy, because of their accuracy and applicability in numerous studies [125, 126].

LDA has been widely used for its fast and simple implementation with low computational requirements. It is suitable for real-time implementation [128]. Its objective is to create a new variable that combines the original predictors by finding a hyperplane that separates the data points representing different classes and that minimizes the variance within the class under the assumption of normal data distribution [125].

SVM is a linear binary classifier that can be used as an alternative to multilayer perceptron. It can support feature vectors with many components [129]. SVM gives the researchers a way to come up with a nonlinear classifier by appropriate kernel methods. Specifically, SVM uses hyperplanes that maximize the distance between the two classes of SVMs based on the principle of maximizing the margin of separation of the classifier to split class [130]. In nonlinear cases, SVM can be extended to the concept of hyperplane separation of data that are often linearly nonseparable. The two classes are mapped with kernel methods onto a new higher dimensional feature space via nonlinear mapping [130]. Figure 12 shows the architecture of the SVM.


SVM is widely used in biomedical signal classification application, particularly in EMG and EEG classifications, for its high accuracy and good performance that make it insensitive to overtraining and dimensionality [126, 132–134]. SVM may help obtain an accurate classification of the severity of dementia that provides an indication of the mental disorder and can predict early stages with suitable treatment management programs.

4. Discussion

EEG plays an important role in evaluating brain activity. This review is focused on using EEG as a physiological biomarker to detect dementia in the early stages and classifying its severity based on EEG signal analysis and processing.

There is enormous interest in the detection and diagnosis of dementia in its early stages. This might be achieved through a combination of diagnosis criteria and reliable biomarkers. The scientific knowledge available through neuropsychological testing and biomarkers assessed against diverse dementia signs would help in capturing both the earliest stages and the spectrum of dementia before significant mental decline [26–29]. There is an urgent need for an accurate, specific, and cost-effective biomarker to diagnose dementia. This makes the EEG an attractive tool to detect and differentiate AD and VaD in the early stages due to its affordability and noninvasiveness. This review has focused on the use of EEG as a physiological biomarker to provide the impetus to detect dementia in the early stages. EEG evaluation through visual inspection is prone to mistakes due to the subjective experience of neurologists. In addition, it is time consuming and it may not be able to reveal subtle changes in the EEG, whereas the computerized EEG signal analysis may simplify the work of medical doctors and may contribute to making the evaluations more objective.

This review illustrated EEG signal processing principles and described useful techniques that have been used to enhance recorded EEG signals. Numerous preprocessing signal and denoising techniques are used to enhance EEG signals by removing artifacts. Methods like WT and ICA have been used to remove different types of noise. On the one hand, ICA, as a higher order statistics method, has several advantages due to its ability to split a set of mixed signals into its sources. Nonetheless, ICA may have difficulties in determining the order of the ICs. However, it is a powerful method for artifact removal and suitable for offline application. On the other hand, WT is suitable for nonstationary signal like EEG that provide linear combination of the sum of wavelet coefficients and mother wavelet with frequency and localization information, and WT has the ability of splitting the signal into subbands (approximation and detail) using a multiresolution decomposition algorithm. In recent years, ICA-wavelet hybrid techniques have been used to overcome the limitation of each individual method and it may become a more effective denoising method. To improve the performance of ICA and WT, the data can be projected into a new space when the redundancy is higher and the features in frequency domain are fully exploited. This minimizes the information loss and it enables WT to remove any overlapping of noise in the EEG signals that ICA cannot filter out.

This review also explored linear and nonlinear features extraction techniques and dimensionality reduction methods. The summary of the findings of the most effective linear and nonlinear methods is listed in Table 1.

Thereafter, the techniques used to classify EEG signals based on dementia spectrum (i.e., CIND, MCI, and dementia) were revised. The effects of dementia on the EEG can be summarized as slowing and reducing EEG complexity and synchrony. The SVM classifier is suggested as a suitable technique for classifying the features of EEG signals based on their applicability in many fields for its empirically good performance and generalization. Many researchers have benefited from the advantages of SVM in dealing with large feature spaces. Other researchers have been applying a combination of classification algorithms that may help improve the performance, sensitivity, and specificity of the best clinical diagnosis for the early detection and classification of dementia.

5. Conclusions

In this review, EEG has been identified as an investigation tool and potential biomarker for detecting dementia and classifying its severity by providing concise information about the brain activity and how it is affected by AD and VaD. It must be noted that, in some occasions, the review has focused on findings related to AD. This is due to the fact that the literature on AD is much larger. Although there has been considerable research into the use of EEG for dementia screening, this is not accepted in routine practice yet [26–29]. Furthermore, the analyzed datasets have often been small and additional studies are needed to confirm those promising results. However, several studies have appreciated the EEG as a useful clinical evaluation tool in the discrimination of AD and/or VaD and/or other types of dementia. Highly sensitive EEG-based detection of the progress of dementia and classification of its severity are a highly desirable screening technique in clinical practice as its low cost and portable features make it a promising technique that can be a reference for customizing or personalizing optimal therapeutic programs for dementia patients.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

References

  1. D. Pond, “Dementia an update on management,” Australian Family Physician, vol. 41, no. 12, pp. 936–939, 2012. View at: Google Scholar
  2. A. Cedazo-Minguez and B. Winblad, “Biomarkers for Alzheimer's disease and other forms of dementia: clinical needs, limitations and future aspects,” Experimental Gerontology, vol. 45, no. 1, pp. 5–14, 2010. View at: Publisher Site | Google Scholar
  3. H. Hampel, R. Frank, K. Broich et al., “Biomarkers for alzheimer's disease: academic, industry and regulatory perspectives,” Nature Reviews Drug Discovery, vol. 9, no. 7, pp. 560–574, 2010. View at: Publisher Site | Google Scholar
  4. J. Jeong, “EEG dynamics in patients with Alzheimer's disease,” Clinical Neurophysiology, vol. 115, no. 7, pp. 1490–1505, 2004. View at: Publisher Site | Google Scholar
  5. E. R. John, L. S. Prichep, J. Fridman, and P. Easton, “Neurometrics: computer-assisted differential diagnosis of brain dysfunctions,” Science, vol. 239, no. 4836, pp. 162–169, 1988. View at: Publisher Site | Google Scholar
  6. A. F. Leuchter, I. A. Cook, T. F. Newton et al., “Regional differences in brain electrical activity in dementia: use of spectral power and spectral ratio measures,” Electroencephalography and Clinical Neurophysiology, vol. 87, no. 6, pp. 385–393, 1993. View at: Publisher Site | Google Scholar
  7. H. Hampel, S. Lista, S. J. Teipel et al., “Perspective on future role of biological markers in clinical therapy trials of Alzheimer's disease: a long-range point of view beyond 2020,” Biochemical Pharmacology, vol. 88, no. 4, pp. 426–449, 2014. View at: Publisher Site | Google Scholar
  8. P. Luu, D. M. Tucker, R. Englander, A. Lockfeld, H. Lutsep, and B. Oken, “Localizing acute stroke-related EEG changes: assessing the effects of spatial undersampling,” Journal of Clinical Neurophysiology, vol. 18, no. 4, pp. 302–317, 2001. View at: Publisher Site | Google Scholar
  9. S. M. Snyder, J. R. Hall, S. L. Cornwell, and J. D. Falk, “Addition of EEG improves accuracy of a logistic model that uses neuropsychological and cardiovascular factors to identify dementia and MCI,” Psychiatry Research, vol. 186, no. 1, pp. 97–102, 2011. View at: Publisher Site | Google Scholar
  10. S. Borson, L. Frank, P. J. Bayley et al., “Improving dementia care: the role of screening and detection of cognitive impairment,” Alzheimer's and Dementia, vol. 9, no. 2, pp. 151–159, 2013. View at: Publisher Site | Google Scholar
  11. S. T. DeKosky and K. Marek, “Looking backward to move forward: early detection of neurodegenerative disorders,” Science, vol. 302, no. 5646, pp. 830–834, 2003. View at: Publisher Site | Google Scholar
  12. P. B. Gorelick, “Status of risk factors for dementia associated with stroke,” Stroke, vol. 28, no. 2, pp. 459–463, 1997. View at: Publisher Site | Google Scholar
  13. J. Jeong, J.-H. Chae, S. Y. Kim, and S.-H. Han, “Nonlinear dynamic analysis of the EEG in patients with Alzheimer's disease and vascular dementia,” Journal of Clinical Neurophysiology, vol. 18, no. 1, pp. 58–67, 2001. View at: Publisher Site | Google Scholar
  14. K. Blennow, M. J. de Leon, and H. Zetterberg, “Alzheimer's disease,” The Lancet, vol. 368, no. 9533, pp. 387–403, 2006. View at: Publisher Site | Google Scholar
  15. E. M. C. Schrijvers, Biomarkers and Risk Factors of Dementia, Erasmus University, Rotterdam, The Netherlands, 2011.
  16. J. T. Coyle, D. L. Price, and M. R. DeLong, “Alzheimer's disease: a disorder of cortical cholinergic innervation,” Science, vol. 219, no. 4589, pp. 1184–1190, 1983. View at: Publisher Site | Google Scholar
  17. A. V. Terry Jr. and J. J. Buccafusco, “The cholinergic hypothesis of age and Alzheimer's disease-related cognitive deficits: recent challenges and their implications for novel drug development,” Journal of Pharmacology and Experimental Therapeutics, vol. 306, no. 3, pp. 821–827, 2003. View at: Publisher Site | Google Scholar
  18. C. E. Jackson and P. J. Snyder, “Electroencephalography and event-related potentials as biomarkers of mild cognitive impairment and mild Alzheimer's disease,” Alzheimer's and Dementia, vol. 4, no. 1, pp. S137–S143, 2008. View at: Publisher Site | Google Scholar
  19. C. McVeigh and P. Passmore, “Vascular dementia: prevention and treatment,” Clinical Interventions in Aging, vol. 1, no. 3, pp. 229–235, 2006. View at: Publisher Site | Google Scholar
  20. A. Ruitenberg, A. Ott, J. C. van Swieten, A. Hofman, and M. M. Breteler, “Incidence of dementia: does gender make a difference?” Neurobiology of Aging, vol. 22, no. 4, pp. 575–580, 2001. View at: Google Scholar
  21. G. C. Román, “Vascular dementia: distinguishing characteristics, treatment, and prevention,” Journal of the American Geriatrics Society, vol. 51, no. 5, supplement 2, pp. S296–S304, 2003. View at: Publisher Site | Google Scholar
  22. G. C. Román, “Vascular dementia may be the most common form of dementia in the elderly,” Journal of the Neurological Sciences, vol. 203-204, pp. 7–10, 2002. View at: Publisher Site | Google Scholar
  23. D. R. Thal, L. T. Grinberg, and J. Attems, “Vascular dementia: different forms of vessel disorders contribute to the development of dementia in the elderly brain,” Experimental Gerontology, vol. 47, no. 11, pp. 816–824, 2012. View at: Publisher Site | Google Scholar
  24. C. Jacova, A. Kertesz, M. Blair, J. D. Fisk, and H. H. Feldman, “Neuropsychological testing and assessment for dementia,” Alzheimer's and Dementia, vol. 3, no. 4, pp. 299–317, 2007. View at: Publisher Site | Google Scholar
  25. R. C. Petersen, “Mild cognitive impairment as a diagnostic entity,” Journal of Internal Medicine, vol. 256, no. 3, pp. 183–194, 2004. View at: Publisher Site | Google Scholar
  26. C. R. Jack Jr., M. S. Albert, D. S. Knopman et al., “Introduction to the recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease,” Alzheimer's and Dementia, vol. 7, no. 3, pp. 257–262, 2011. View at: Publisher Site | Google Scholar
  27. M. S. Albert, S. T. DeKosky, D. Dickson et al., “The diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease,” Alzheimer's and Dementia, vol. 7, no. 3, pp. 270–279, 2011. View at: Publisher Site | Google Scholar
  28. G. M. McKhann, D. S. Knopman, H. Chertkow et al., “The diagnosis of dementia due to Alzheimer's disease: Recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease,” Alzheimer's and Dementia, vol. 7, no. 3, pp. 263–269, 2011. View at: Publisher Site | Google Scholar
  29. R. A. Sperling, P. S. Aisen, L. A. Beckett et al., “Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease,” Alzheimer's and Dementia, vol. 7, no. 3, pp. 280–292, 2011. View at: Publisher Site | Google Scholar
  30. G. McKhann, D. Drachman, and M. Folstein, “Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA work group * under the auspices of Department of Health and Human Services Task Force on Alzheimer's disease,” Neurology, vol. 34, no. 7, pp. 939–944, 1984. View at: Publisher Site | Google Scholar
  31. B. Sheng, L. F. Cheng, C. B. Law, H. L. Li, K. M. Yeung, and K. K. Lau, “Coexisting cerebral infarction in Alzheimer's disease is associated with fast dementia progression: applying the National Institute for Neurological Disorders and Stroke/Association Internationale pour la Recherche et l'Enseignement en Neurosciences neuroimaging criteria in Alzheimer's disease with concomitant cerebral infarction,” Journal of the American Geriatrics Society, vol. 55, no. 6, pp. 918–922, 2007. View at: Publisher Site | Google Scholar
  32. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, American Psychiatric Association, Washington, DC, USA, 4th edition, 1994.
  33. C. P. Hughes, L. Berg, and W. L. Danziger, “A new clinical scale for the staging of dementia,” The British Journal of Psychiatry, vol. 140, no. 6, pp. 566–572, 1982. View at: Publisher Site | Google Scholar
  34. J. A. Yesavage, T. L. Brink, T. L. Rose et al., “Development and validation of a geriatric depression screening scale: A preliminary report,” Journal of Psychiatric Research, vol. 17, no. 1, pp. 37–49, 1982. View at: Publisher Site | Google Scholar
  35. W. G. Rosen, R. D. Terry, P. A. Fuld, R. Katzman, and A. Peck, “Pathological verification of ischemic score in differentiation of dementias,” Annals of Neurology, vol. 7, no. 5, pp. 486–488, 1980. View at: Publisher Site | Google Scholar
  36. M. P. Lawton and E. M. Brody, “Assessment of older people: self-maintaining and instrumental activities of daily living,” The Gerontologist, vol. 19, no. 3, pp. 179–186, 1969. View at: Google Scholar
  37. M. F. Folstein, S. E. Folstein, and P. R. McHugh, “Mini-mental state,” A prac-32, 1998. View at: Google Scholar
  38. T. Smith, N. Gildeh, and C. Holmes, “The Montreal cognitive assessment: validity and utility in a memory clinic setting,” Canadian Journal of Psychiatry, vol. 52, no. 5, pp. 329–332, 2007. View at: Google Scholar
  39. P. S. Mathuranath, P. J. Nestor, G. E. Berrios, W. Rakowicz, and J. R. Hodges, “A brief cognitive test battery to differentiate Alzheimer's disease and frontotemporal dementia,” Neurology, vol. 55, no. 11, pp. 1613–1620, 2000. View at: Publisher Site | Google Scholar
  40. P. Amodio, H. Wenin, F. Del Piccolo et al., “Variability of trail making test, symbol digit test and line trait test in normal people. A normative study taking into account age-dependent decline and sociobiological variables,” Aging: Clinical and Experimental Research, vol. 14, no. 2, pp. 117–131, 2002. View at: Publisher Site | Google Scholar
  41. K. I. Shulman, “Clock-drawing: is it the ideal cognitive screening test?” International Journal of Geriatric Psychiatry, vol. 15, no. 6, pp. 548–561. View at: Google Scholar
  42. P. Caffarra, G. Vezzadini, F. Dieci, F. Zonato, and A. Venneri, “Rey-Osterrieth complex figure: normative values in an Italian population sample,” Neurological Sciences, vol. 22, no. 6, pp. 443–447, 2002. View at: Publisher Site | Google Scholar
  43. G. A. Carlesimo, C. Caltagirone, G. Gainotti et al., “The mental deterioration battery: normative data, diagnostic reliability and qualitative analyses of cognitive impairment,” European Neurology, vol. 36, no. 6, pp. 378–384, 1996. View at: Publisher Site | Google Scholar
  44. V. Dorval, P. T. Nelson, and S. S. Hrt, “Circulating microRNAs in Alzheimer's disease: the search for novel biomarkers,” Frontiers in Molecular Neuroscience, vol. 6, article 24, 2013. View at: Google Scholar
  45. J. A. Sonnen, K. S. Montine, J. F. Quinn, J. A. Kaye, J. C. Breitner, and T. J. Montine, “Biomarkers for cognitive impairment and dementia in elderly people,” The Lancet Neurology, vol. 7, no. 8, pp. 704–714, 2008. View at: Publisher Site | Google Scholar
  46. S.-S. Poil, W. de Haan, W. M. van der, H. D. Flier, P. Scheltens, and K. Linkenkaer-Hansen, “Integrative EEG biomarkers predict progression to Alzheimer's disease at the MCI stage,” Frontiers in Aging Neuroscience, vol. 5, article 58, 2013. View at: Google Scholar
  47. N. Mattsson, H. Zetterberg, O. Hansson et al., “CSF biomarkers and incipient Alzheimer disease in patients with mild cognitive impairment,” JAMA: Journal of the American Medical Association, vol. 302, no. 4, pp. 385–393, 2009. View at: Publisher Site | Google Scholar
  48. G. P. Paraskevas, E. Kapaki, S. G. Papageorgiou et al., “CSF biomarker profile and diagnostic value in vascular dementia,” European Journal of Neurology, vol. 16, no. 2, pp. 205–211, 2009. View at: Publisher Site | Google Scholar
  49. S. V. Frankfort, L. R. Tulner, J. P. C. M. van Campen, M. M. Verbeekl, R. W. M. M. Jansen, and J. H. Beijnen, “Amyloid beta protein and tau in cerebrospinal fluid and plasma as biomarkers for dementia: a review of recent literature,” Current Clinical Pharmacology, vol. 3, no. 2, pp. 123–131, 2008. View at: Publisher Site | Google Scholar
  50. M. Folin, S. Baiguera, M. T. Conconi et al., “Apolipoprotein E as vascular risk factor in neurodegenerative dementia,” International Journal of Molecular Medicine, vol. 14, no. 4, pp. 609–613, 2004. View at: Google Scholar
  51. P. M. Rossini, S. Rossi, C. Babiloni, and J. Polich, “Clinical neurophysiology of aging brain: from normal aging to neurodegeneration,” Progress in Neurobiology, vol. 83, no. 6, pp. 375–400, 2007. View at: Publisher Site | Google Scholar
  52. A. L. Schneider and K. G. Jordan, “Regional wttenuation without Delta (RAWOD): a distinctive EEG pattern that can aid in the diagnosis and management of severe acute ischemic stroke,” The American Journal of Electroneurodiagnostic Technology, vol. 45, no. 2, pp. 102–117, 2005. View at: Google Scholar
  53. P. R. Davidson, R. D. Jones, and M. T. Peiris, “EEG-based lapse detection with high temporal resolution,” IEEE Transactions on Biomedical Engineering, vol. 54, no. 5, pp. 832–839, 2007. View at: Publisher Site | Google Scholar
  54. F. Vecchio, C. Babiloni, R. Lizio et al., “Resting state cortical EEG rhythms in Alzheimer's disease: toward EEG markers for clinical applications: a review,” Supplements to Clinical Neurophysiology, vol. 62, pp. 223–236, 2013. View at: Publisher Site | Google Scholar
  55. G. Henderson, E. Ifeachor, N. Hudson et al., “Development and assessment of methods for detecting dementia using the human electroencephalogram,” IEEE Transactions on Biomedical Engineering, vol. 53, no. 8, pp. 1557–1568, 2006. View at: Publisher Site | Google Scholar
  56. J. J. Claus, B. W. Ongerboer de Visser, G. J. M. Walstra, A. Hijdra, B. Verbeeten Jr., and W. A. van Gool, “Quantitative spectral electroencephalography in predicting survival in patients with early Alzheimer disease,” Archives of Neurology, vol. 55, no. 8, pp. 1105–1111, 1998. View at: Publisher Site | Google Scholar
  57. J. J. Claus, B. W. Ongerboer de Visser, L. J. Bour et al., “Determinants of quantitative spectral electroencephalography in early Alzheimer's disease: cognitive function, regional cerebral blood flow, and computed tomography,” Dementia and Geriatric Cognitive Disorders, vol. 11, no. 2, pp. 81–89, 2000. View at: Publisher Site | Google Scholar
  58. E. Helkala, V. Laulumaa, R. Soikkeli, J. Partanen, H. Soininen, and P. J. Riekkinen, “Slow-wave activity in the spectral analysis of the electroencephalogram is associated with cortical dysfunctions in patients with Alzheimer's disease,” Behavioral Neuroscience, vol. 105, no. 3, pp. 409–415, 1991. View at: Publisher Site | Google Scholar
  59. A. A. Petrosian, D. V. Prokhorov, W. Lajara-Nanson, and R. B. Schiffer, “Recurrent neural network-based approach for early recognition of Alzheimer's disease in EEG,” Clinical Neurophysiology, vol. 112, no. 8, pp. 1378–1387, 2001. View at: Publisher Site | Google Scholar
  60. G. Henderson, E. Ifeachor, H. Wimalaratna, T. Allen, and R. Hudson, “Prospects for routine detection of dementia using the fractal dimension of the human electroencephalogram,” IEE Proceedings—Science Measurement and Technology, vol. 147, no. 6, pp. 321–326, 2000. View at: Publisher Site | Google Scholar
  61. P. Zhao and E. Ifeachor, “EEG assessment of Alzheimers diseases using universal compression algorithm”. View at: Google Scholar
  62. J. G. Webster, Medical Instrumentation Application and Design, Wiley, New York, NY, USA, 4th edition, 1998.
  63. E. J. Austin, R. J. Wilkus, and W. T. Longstreth Jr., “Etiology and prognosis of alpha coma,” Neurology, vol. 38, no. 5, pp. 773–777, 1988. View at: Publisher Site | Google Scholar
  64. D. V. Moretti, O. Zanetti, G. Binetti, and G. B. Frisoni, “Quantitative EEG markers in mild cognitive impairment: degenerative versus vascular brain impairment,” International Journal of Alzheimer's Disease, vol. 2012, Article ID 917537, 12 pages, 2012. View at: Publisher Site | Google Scholar
  65. D. V. Moretti, C. Miniussi, G. Frisoni et al., “Vascular damage and EEG markers in subjects with mild cognitive impairment,” Clinical Neurophysiology, vol. 118, no. 8, pp. 1866–1876, 2007. View at: Publisher Site | Google Scholar
  66. D. A. Pizzagalli, “Electroencephalography and high-density electrophysiological source localization,” in Handbook of Psychophysiology, pp. 56–84, Cambridge University Press, New York, NY, USA, 3rd edition, 2007. View at: Publisher Site | Google Scholar
  67. E. R. John, H. Ahn, L. Prichep, M. Trepetin, D. Brown, and H. Kaye, “Developmental equations for the electroencephalogram,” Science, vol. 210, no. 4475, pp. 1255–1258, 1980. View at: Publisher Site | Google Scholar
  68. T. R. Oakes, D. A. Pizzagalli, A. M. Hendrick et al., “Functional coupling of simultaneous electrical and metabolic activity in the human brain,” Human Brain Mapping, vol. 21, no. 4, pp. 257–270, 2004. View at: Publisher Site | Google Scholar
  69. B. Saletu, P. Anderer, E. Paulus et al., “EEG brain mapping in diagnostic and therapeutic assessment of dementia,” Alzheimer Disease and Associated Disorders, vol. 5, no. 1, pp. S57–S75, 1991. View at: Publisher Site | Google Scholar
  70. M. Signorino, E. Pucci, N. Belardinelli, G. Nolfe, and F. Angeleri, “EEG spectral analysis in vascular and Alzheimer dementia,” Electroencephalography and Clinical Neurophysiology, vol. 94, no. 5, pp. 313–325, 1995. View at: Publisher Site | Google Scholar
  71. E. Pucci, G. Cacchiò, R. Angeloni et al., “EEG spectral analysis in Alzheimer's disease and different degenerative dementias,” Archives of Gerontology and Geriatrics, vol. 26, no. 3, pp. 283–297, 1998. View at: Publisher Site | Google Scholar
  72. E. B. Gordon and M. Sim, “The E.E.G. in presenile dementia,” Journal of Neurology Neurosurgery and Psychiatry, vol. 30, no. 3, pp. 285–291, 1967. View at: Publisher Site | Google Scholar
  73. K. J. Stam, D. L. J. Tavy, B. Jelles, H. A. M. Achtereekte, J. P. J. Slaets, and R. W. M. Keunen, “Non-linear dynamical analysis of multichannel EEG: clinical applications in dementia and Parkinson's disease,” Brain Topography, vol. 7, no. 2, pp. 141–150, 1994. View at: Publisher Site | Google Scholar
  74. H. Berger, “�r das elektrenkephalogramm des menschen,” Archiv für Psychiatrie und Nervenkrankheiten, vol. 87, no. 1, pp. 527–570, 1929. View at: Publisher Site | Google Scholar
  75. H. Berger, “�r das elektrenkephalogramm des menschen,” European Archives of Psychiatry and Clinical Neuroscience, vol. 98, no. 1, pp. 231–254, 1933. View at: Publisher Site | Google Scholar
  76. I. Rosén, “Electroencephalography as a diagnostic tool in dementia,” Dementia and Geriatric Cognitive Disorders, vol. 8, no. 2, pp. 110–116, 1997. View at: Publisher Site | Google Scholar
  77. D. V. Moretti, G. B. Frisoni, C. Fracassi et al., “MCI patients' EEGs show group differences between those who progress and those who do not progress to AD,” Neurobiology of Aging, vol. 32, no. 4, pp. 563–571, 2011. View at: Publisher Site | Google Scholar
  78. D. Moretti, D. Paternicò, G. Binetti, O. Zanetti, and G. Frisoni, “Relationship between EEG alpha3/alpha2 ration and the nucleus accumbens in subjects with mild cognitive impairment,” Journal of Neurology & Neurophysiology, vol. 4, pp. 1–6, 2013. View at: Google Scholar
  79. D. V. Moretti, A. Prestia, G. Binetti, O. Zanetti, and G. B. Frisoni, “Increase of theta frequency is associated with reduction in regional cerebral blood flow only in subjects with mild cognitive impairment with higher upper alpha/low alpha EEG frequency power ratio,” Frontiers in Behavioral Neuroscience, vol. 7, article 188, 2013. View at: Publisher Site | Google Scholar
  80. S. A. Taywade and R. D. Raut, “A review: EEG signal analysis with different methodologies,” in Proceedings of the National Conference on Innovative Paradigms in Engineering and Technology (NCIPET '12), pp. 29–31, 2014. View at: Google Scholar
  81. R. S. Khandpur, Handbook of Biomedical Instrumentation, Tata McGraw-Hill, New Delhi, India, 1998.
  82. W. O. Tatum, A. M. Husain, S. R. Benbadis, and P. W. Kaplan, Handbook of EEG Interpretation, Demos Medical Publishing, LLC, New York, NY, USA, 2008.
  83. S. Chatteriee and A. Miller, Biomedical Instrumentation System, Cengage Learning, Delmar, Del, USA, 2010.
  84. B. Hamadicharef, C. Guan, N. Hudson, E. C. Ifeachor, and S. Wimalaratna, “Performance evaluation and fusion of methods for early detection of Alzheimer Disease,” in Proceedings of the 1st International Conference on BioMedical Engineering and Informatics (BMEI ✈), pp. 347–351, Sanya, China, May 2008. View at: Publisher Site | Google Scholar
  85. S. Sanei and J. A. Chambers, EEG Signal Procesing, John Wiley & Sons, New York, NY, USA, 2007.
  86. D. V. Moretti, C. Babiloni, G. Binetti et al., “Individual analysis of EEG frequency and band power in mild Alzheimer's disease,” Clinical Neurophysiology, vol. 115, no. 2, pp. 299–308, 2004. View at: Publisher Site | Google Scholar
  87. T. Jung, S. Makeig, M. Westerfield, J. Townsend, E. Courchesne, and T. J. Sejnowski, “Removal of eye activity artifacts from visual event-related potentials in normal and clinical subjects,” Clinical Neurophysiology, vol. 111, no. 10, pp. 1745–1758, 2000. View at: Publisher Site | Google Scholar
  88. M. Habl, C. Bauer, C. Ziegaus, E. Lang, and F. Schulmeyer, “Can ICA help identify brain tumor related EEG signals,” pp. 609-614, http://users.ics.aalto.fi/ica2000/proceedings/0609.pdf. View at: Google Scholar
  89. C. Guerrero-Mosquera, A. M. Trigueros, and A. Navia-Vazquez, “EEG signal processing for epilepsy,” Epilepsy–Histological, Electroencephalographic and Psychological Aspects, vol. 3, pp. 49–74, 2012. View at: Google Scholar
  90. I. M. B. Nú༞z, EEG artifact dtection, 2010.
  91. G. N. G. Molina, “Direct brain-computer communication through scalp recorded EEG signals,” ಜole Polytechnique Fຝérale de Lausanne, 2004. View at: Google Scholar
  92. A. Naït-Ali, Advanced Biosignal Processing, Springer, New York, NY, USA, 2009.
  93. D. Langlois, S. Chartier, and D. Gosselin, “An introduction to independent component analysis: InfoMax and FastICA algorithms,” Tutorials in Quantitative Methods for Psychology, vol. 6, no. 1, pp. 31–38, 2010. View at: Google Scholar
  94. M. J. McKeown, C. Humphries, P. Achermann, A. A. Borbely, and T. J. Sejnowski, “A new method for detecting state changes in the EEG: exploratory application to sleep data,” Journal of Sleep Research, vol. 7, no. S1, pp. 48–56, 1998. View at: Publisher Site | Google Scholar
  95. T. Zikov, S. Bibian, G. A. Dumont, M. Huzmezan, and C. R. Ries, “A wavelet based de-noising technique for ocular artifact correction of the electroencephalogram,” in Proceedings of the IEEE Engineering in Medicine and Biology 24th Annual Conference and the Fall Meeting of the Biomedical Engineering Society, pp. 98–105, October 2002. View at: Google Scholar
  96. V. Krishnaveni, S. Jayaraman, S. Aravind, V. Hariharasudhan, and K. Ramadoss, “Automatic identification and removal of ocular artifacts from EEG using wavelet transform,” Measurement Science Review, vol. 6, no. 4, pp. 45–57, 2006. View at: Google Scholar
  97. P. S. Kumar, R. Arumuganathan, K. Sivakumar, and C. Vimal, “Removal of ocular artifacts in the EEG through wavelet transform without using an EOG reference channel,” International Journal of Open Problems in Computer Science and Mathematics, vol. 1, no. 3, pp. 188–200, 2008. View at: Google Scholar
  98. V. Krishnaveni, S. Jayaraman, L. Anitha, and K. Ramadoss, “Removal of ocular artifacts from EEG using adaptive thresholding of wavelet coefficients,” Journal of Neural Engineering, vol. 3, no. 4, article 11, p. 338, 2006. View at: Publisher Site | Google Scholar
  99. P. Jahankhani, V. Kodogiannis, and K. Revett, “EEG signal classification using wavelet feature extraction and neural networks,” in Proceedings of the IEEE John Vincent Atanasoff International Symposium on Modern Computing (JVA '06), pp. 120–124, Sofia, Bulgaria, October 2006. View at: Publisher Site | Google Scholar
  100. M. T. Akhtar and C. J. James, “Focal artifact removal from ongoing EEG𠅊 hybrid approach based on spatially-constrained ICA and wavelet de-noising,” in Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC ✉), pp. 4027–4030, Minneapolis, Minn, USA, September 2009. View at: Publisher Site | Google Scholar
  101. M. T. Akhtar, C. J. James, and W. Mitsuhashi, “Modifying the spatially-constrained ICA for efficient removal of artifacts from EEG data,” in Proceedings of the 4th International Conference on Bioinformatics and Biomedical Engineering (iCBBE '10), pp. 1–4, Chengdu, China, June 2010. View at: Publisher Site | Google Scholar
  102. J. Walters-Williams and Y. Li, “A new approach to denoising EEG signals-merger of translation invariant wavelet and ICA,” International Journal of Biometrics and Bioinformatics, vol. 5, no. 2, pp. 130–149, 2011. View at: Google Scholar
  103. J. Walters-Williams and Y. Li, “Performance comparison of known ICA algorithms to a wavelet-ICA merger,” Signal Processing, vol. 5, no. 3, p. 80, 2011. View at: Google Scholar
  104. N. Mammone, F. La Foresta, and F. C. Morabito, “Automatic artifact rejection from multichannel scalp EEG by wavelet ICA,” IEEE Sensors Journal, vol. 12, no. 3, pp. 533–542, 2012. View at: Publisher Site | Google Scholar
  105. G. Inuso, F. La Foresta, N. Mammone, and F. C. Morabito, “Wavelet-ICA methodology for efficient artifact removal from Electroencephalographic recordings,” in Proceedings of the International Joint Conference on Neural Networks (IJCNN ✇), pp. 1524–1529, Orlando, Fla, USA, August 2007. View at: Publisher Site | Google Scholar
  106. B. Jelles, P. Scheltens, W. M. van der Flier, E. J. Jonkman, F. H. L. da Silva, and C. J. Stam, “Global dynamical analysis of the EEG in Alzheimer's disease: Frequency-specific changes of functional interactions,” Clinical Neurophysiology, vol. 119, no. 4, pp. 837–841, 2008. View at: Publisher Site | Google Scholar
  107. J. Escudero, R. Hornero, D. Abásolo, and A. Fernández, “Blind source separation to enhance spectral and non-linear features of magnetoencephalogram recordings. Application to Alzheimer's disease,” Medical Engineering & Physics, vol. 31, no. 7, pp. 872–879, 2009. View at: Publisher Site | Google Scholar
  108. R. Hornero, J. Escudero, A. Fernández, J. Poza, and C. Gómez, “Spectral and nonlinear analyses of MEG background activity in patients with Alzheimer's disease,” IEEE Transactions on Biomedical Engineering, vol. 55, no. 6, pp. 1658–1665, 2008. View at: Publisher Site | Google Scholar
  109. O. N. Markand, “Organic brain syndromes and dementias,” Current Practice of Clinical Electroencephalography, vol. 3, pp. 378–404, 1990. View at: Google Scholar
  110. J. Dauwels, F.-B. Vialatte, and A. Cichocki, “On the early diagnosis of Alzheimer's disease from EEG signals: a mini-review,” in Advances in Cognitive Neurodynamics II, pp. 709–716, Springer, 2011. View at: Google Scholar
  111. J. Escudero, S. Sanei, D. Jarchi, D. Abásolo, and R. Hornero, “Regional coherence evaluation in mild cognitive impairment and Alzheimer's disease based on adaptively extracted magnetoencephalogram rhythms,” Physiological Measurement, vol. 32, no. 8, pp. 1163–1180, 2011. View at: Publisher Site | Google Scholar
  112. J. Jeong, “Nonlinear dynamics of EEG in Alzheimer's disease,” Drug Development Research, vol. 56, no. 2, pp. 57–66, 2002. View at: Publisher Site | Google Scholar
  113. D. P. Subha, P. K. Joseph, R. Acharya, and C. M. Lim, “EEG signal analysis: a survey,” Journal of Medical Systems, vol. 34, no. 2, pp. 195–212, 2010. View at: Publisher Site | Google Scholar
  114. D. Abásolo, R. Hornero, C. Gómez, M. Garc໚, and M. López, “Analysis of EEG background activity in Alzheimer's disease patients with Lempel-Ziv complexity and central tendency measure,” Medical Engineering and Physics, vol. 28, no. 4, pp. 315–322, 2006. View at: Publisher Site | Google Scholar
  115. J. Escudero, D. Abásolo, R. Hornero, P. Espino, and M. López, “Analysis of electroencephalograms in Alzheimer's disease patients with multiscale entropy,” Physiological Measurement, vol. 27, no. 11, article 004, p. 1091, 2006. View at: Publisher Site | Google Scholar
  116. P. Grassberger and I. Procaccia, “Measuring the strangeness of strange attractors,” Physica D: Nonlinear Phenomena, vol. 9, no. 1-2, pp. 189–208, 1983. View at: Publisher Site | Google Scholar | MathSciNet
  117. A. Wolf, J. B. Swift, and H. L. a. Swinney, “Determining Lyapunov exponents from a time series,” Physica D: Nonlinear Phenomena, vol. 16, no. 3, pp. 285–317, 1985. View at: Publisher Site | Google Scholar | MathSciNet
  118. B. Hamadicharef, C. Guan, E. C. Ifeachor, N. Hudson, E. C. Ifeachor, and S. Wimalaratna, “Performance evaluation and fusion of methods for early detection of Alzheimer Disease,” in Proceedings of the 1st International Conference on BioMedical Engineering and Informatics (BMEI '08), pp. 347–351, May 2008. View at: Publisher Site | Google Scholar
  119. G. T. Henderson, Early Detection of Dementia Using The Human Electroencephalogram, University of Plymouth, 2004.
  120. R. Ferenets, T. Lipping, A. Anier, V. Jantti, S. Melto, and S. Hovilehto, “Comparison of entropy and complexity measures for the assessment of depth of sedation,” IEEE Transactions on Biomedical Engineering, vol. 53, no. 6, pp. 1067–1077, 2006. View at: Publisher Site | Google Scholar
  121. P. Zhao, P. Van-Eetvelt, C. Goh, N. Hudson, S. Wimalaratna, and E. Ifeachor, “Characterization of EEGs in Alzheimer's disease using information theoretic methods,” in Proceedings of the 29th Annual International Conference of the IEEEon Engineering in Medicine and Biology Society (EMBC ✇), pp. 5127–5131, Lyon, France, August 2007. View at: Publisher Site | Google Scholar
  122. P. Zhao, P. Van-Eetvelt, C. Goh, N. Hudson, S. Wimalaratna, and E. Ifeachor, “Characterization of EEGs in alzheimer's disease using information theoretic methods,” in Proceedings of the 29th Annual International Conference of IEEE-EMBS, Engineering in Medicine and Biology Society (EMBC '07), pp. 5127–5131, Lyon , France, August 2007. View at: Publisher Site | Google Scholar
  123. M. Costa, A. L. Goldberger, and C.-K. Peng, “Multiscale entropy analysis of complex physiologic time series,” Physical Review Letters, vol. 89, no. 6, Article ID 068102, 2002. View at: Google Scholar
  124. M. Costa, A. L. Goldberger, and C. Peng, “Multiscale entropy analysis of biological signals,” Physical Review E, vol. 71, no. 2, Article ID 021906, 2005. View at: Publisher Site | Google Scholar | MathSciNet
  125. A. Subasi and M. I. Gursoy, “EEG signal classification using PCA, ICA, LDA and support vector machines,” Expert Systems with Applications, vol. 37, no. 12, pp. 8659–8666, 2010. View at: Publisher Site | Google Scholar
  126. M. KavitaMahajan and M. S. M. Rajput, “A Comparative study of ANN and SVM for EEG classification,” International Journal of Engineering, vol. 1, no. 6, 2012. View at: Google Scholar
  127. F. Vialatte, A. Cichocki, G. Dreyfus, T. Musha, T. M. Rutkowski, and R. Gervais, “Blind source separation and sparse bump modelling of time frequency representation of EEG signals: new tools for early detection of Alzheimer's disease,” in Proceedings of the IEEE Workshop on Machine Learning for Signal Processing, pp. 27–32, Mystic, Conn, USA, September 2005. View at: Publisher Site | Google Scholar
  128. M. Besserve, K. Jerbi, F. Laurent, S. Baillet, J. Martinerie, and L. Garnero, “Classification methods for ongoing EEG and MEG signals,” Biological Research, vol. 40, no. 4, pp. 415–437, 2007. View at: Google Scholar
  129. G. Costantini, D. Casali, and M. Todisco, “An SVM based classification method for EEG signals,” in Proceedings of the 14th WSEAS International Conference on Circuits (ICC '10), pp. 107–109, July 2010. View at: Google Scholar
  130. D. Garrett, D. A. Peterson, C. W. Anderson, and M. H. Thaut, “Comparison of linear, nonlinear, and feature selection methods for EEG signal classification,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 11, no. 2, pp. 141–144, 2003. View at: Publisher Site | Google Scholar
  131. E. C. Ifeachor, N. J. Outram, G. T. Henderson et al., “Nonlinear methods for biopattern analysis: Role and challenges,” in Proceedings of the 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC '04), pp. 5400–5406, September 2004. View at: Google Scholar
  132. C. Lehmann, T. Koenig, V. Jelic et al., “Application and comparison of classification algorithms for recognition of Alzheimer's disease in electrical brain activity (EEG),” Journal of Neuroscience Methods, vol. 161, no. 2, pp. 342–350, 2007. View at: Publisher Site | Google Scholar
  133. R. H. Chowdhury, M. B. Reaz, M. A. B. M. Ali, A. A. Bakar, K. Chellappan, and T. Chang, “Surface electromyography signal processing and classification techniques,” Sensors, vol. 13, no. 9, pp. 12431–12466, 2013. View at: Publisher Site | Google Scholar
  134. A. Al-Timemy, G. Bugmann, J. Escudero, and N. Outram, “Classification of finger movements for the dexterous hand prosthesis control with surface electromyography,” IEEE Journal of Biomedical and Health Informatics, vol. 17, no. 3, pp. 608–618, 2013. View at: Publisher Site | Google Scholar

Copyright

Copyright © 2014 Noor Kamal Al-Qazzaz et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


MEA Technology

This chapter reviews the technology involved in MEA development.

Device Types and Terminology

Over the years, a wide repertoire of terms has been used to refer to and distinguish between all the different forms of MEAs, e.g., emphasizing the type of transducers used (multi-transistor array, microelectrode array, multielectrode array, micro-nail array, capacitive-coupled array, 3D MEA), the type of substrate (active array, passive array, silicon array, CMOS array), the shape of the device (needle-type probe, polytrode, neuro dish), the channel count (multichannel array), the electrode density (HDMEA) or the application (implantable array, in vivo MEA, in vitro MEA) and more. We would therefore like to briefly explain the terminology used in the context of this review. We generalize the term microelectrodes and MEA to cover both substrate-integrated planar MEAs and implantable neural probes. We also include capacitive-coupled devices, such as multi-transistor arrays in the definition of MEAs. We then distinguish between implantable, in vivo MEAs, such as polytrodes and neural probes, and in vitro MEAs that generally include a cell culture dish or some other sort of medium chamber. Further, we classify the different array architectures, as will be explained in Section Advances in MEA and Probe Devices (Figure 3). Briefly, we distinguish between 𠇏ixed wiring” arrays, meaning that each transducer in the array has a direct wire to the outside of the array and “multiplexed arrays,” in which some sort of switching mechanism is employed within the array. We use the term 𠇊rray” to refer to the actual area that encompasses the transducer elements only and we use device or MEA to refer to the entire device. With system, we refer to the MEA and all required components to operate it, such as the data acquisition hardware and software. We use the terms �tive” and “passive” to distinguish between devices with active circuit elements, such as transistors, and devices without such elements.

Electrodes and Transducers

There are various techniques for fabricating microelectrodes, which are reviewed by Li et al. (2003), Park and Shuler (2003), Huang et al. (2009). Choosing the materials for the insulator, conductor, microelectrode, and substrate is crucial, in particular with respect to biocompatibility. All materials in the MEA that will be near to or in contact with cells and tissue need to be tested for toxicity in prolonged periods of time (Hassler et al., 2011). It is also important to consider the biological experiments for which the microelectrodes will be used, whether in vivo or in vitro, culture or acute preparation. Moreover, deciding the type of MEA to use is highly dependent on the type of recorded signals needed, whether EAPs and/or LFPs or intracellular action potentials (IAPs), single cell resolution or not. If the MEA is to be used for stimulation, the charge capacity of electrodes is an important aspect. The electrode needs to be able to mediate reactions at the electrode-electrolyte interface to allow electron flow in the electrode to transition into ion flow in the electrolyte toward stimulating nearby cells (Cogan, 2008).

Generally, an important goal of electrode fabrication is to achieve low impedance. Low electrode impedance results in higher signal-to-noise ratio (SNR), with the usual target SNR of 5:1 or higher. Uniformity of the electrode impedance across an array of electrodes may also be important to obtain consistent data.

Typically, electrodes are made with metallic conductors such as gold (Au), titanium nitride (TiN), platinum (Pt), stainless steel, aluminum (Al), and alloys like iridium oxide (IrOx). Since the electrodes used in MEAs are on the micrometer scale, it is a challenge to achieve low electrode impedance with plain conductors only. Increasing the effective surface area of electrodes can be achieved by modification with porous conductive materials such as Pt-black, Au nanostructures, carbon nanotubes (CNTs), and conductive polymers like poly(3,4-ethylenedioxythiophene) (PEDOT). Emerging materials aside from PEDOT and CNTs include doped diamond and graphene. By modifying the surface, the electrode impedance can be decreased drastically and neuronal recording can be improved (Cui et al., 2001 Franks et al., 2005 Ludwig et al., 2006 Keefer et al., 2008 Viswam et al., 2014). Nam and Wheeler (2011), Kim et al. (2014) for a review of electrode materials and surface modification.

Non-metallic electrodes have been mostly used in conjunction with field-effect transistor (FET) based transducers (Bergveld, 1970 Fromherz et al., 1991). An OGFET can, e.g., be obtained if the fabrication process of a FET is stopped before depositing the gate material (Jenkner et al., 2004). Easier to fabricate is the so-called extended-gate FET (EGFET), in which the FET is fabricated without modification from a standard CMOS process. Metal and via interconnections are used to extend the gate to the surface of the chip, where an insulated electrode implements the 𠇎xtended gate.” Such insulation ensures that no faradaic currents occur. However, as Hierlemann et al., pointed out, devices with metal electrodes also usually connect to a FET directly (Imfeld et al., 2008) or through a filter capacitor (Heer et al., 2006), resulting in a largely capacitive recording situation (Hierlemann et al., 2011). OGFET, EGFET, and devices that directly connect the electrode to the first FET usually need to include some measures to properly bias the gate or some calibration mechanism, which may cause transient currents to flow at the electrode. Whereas for devices with a capacitively coupled front-end stage, controlling the electrode input node is generally not needed. Devices with a FET-based transducer, but using a metalized gate exposed to the liquid, have also been developed (Jobling et al., 1981).

Recently, ultra-small electrodes are being developed to record intracellular activity, including subthreshold signals, as reviewed in Spira and Hai (2013). This is achieved by 3D structured electrodes such as silicon nanowires (Robinson et al., 2013) and Au mushrooms (Hai et al., 2009) penetrating the cell membrane. Electroporation was shown to facilitate measurement of intracellular activity (Koester et al., 2010 Hai and Spira, 2012).

Advances in MEA and Probe Devices

Since the single extracellular microelectrodes used in the middle of the last century (Weale, 1951 Gesteland et al., 1959), development quickly proceeded to MEAs with multiple transducers for the purpose of increasing the number of neurons observed (Thomas et al., 1972 Gross et al., 1977 Pine, 1980 Csicsvari et al., 2003) to increase reliability of spike sorting (Gray et al., 1995 Harris et al., 2000) and to allow for source localization (Blanche et al., 2005 Chelaru and Jog, 2005 Frey et al., 2009b Somogyvári et al., 2012 Delgado Ruz and Schultz, 2014). The advances in lithographic techniques, fueled by the semiconductor industry, allowed a gradual increase in performance and reliability of such multichannel devices. Passive transducer devices based on electrodes embedded in glass or silicon substrates with fixed wiring to amplifiers for in vitro and also in vivo applications became commercially available in the late 90 s and early years of this century. Already early on, silicon-based biosensors for interfacing cells with microelectronics were developed (Bergveld, 1970 Parce et al., 1989). Active devices, employing FETs were fabricated and 2D arrays demonstrated (Besl and Fromherz, 2002). Devices using CMOS technology were fabricated in academic facilities (DeBusschere and Kovacs, 2001) and industrial foundries, usually in conjunction with additional processing steps for biocompatibility reasons (Berdondini et al., 2002 Eversmann et al., 2003b Franks et al., 2003).

The key advantage of integrating active electronic components on the same substrate as the actual electrodes is the possibility of a much higher electrode number and density. Due to the possibility of using active switches to time multiplex signals, integrated circuits make it feasible to transfer data from such high channel counts off chip and to overcome the connectivity limitation of passive devices. Additionally, such co-integration allows amplifying the signals with optimal quality, due to minimal parasitic capacitances and resistances (Hierlemann et al., 2011). The monolithic co-integration also allows including additional functionality, e.g., on-chip spike detection, closed-loop capabilities, electrical stimulation, electronic chip identification, device calibration, and other type of sensing modalities, such as temperature, pH or optical sensing (Baumann et al., 1999 Tokuda et al., 2006 Johnson et al., 2013b).

Figure 2A compares a variety of historical and current devices, to illustrate the evolution of MEAs with respect to overall sensing area and electrode densities. The electrode count is shown with solid lines. The devices are categorized into fixed wiring (Type Aɫ in Figure 3) and multiplexed arrays (Types C𠄾 in Figure 3). Fixed-wiring arrays include devices without any on-chip circuitry (Alpha MED Science Co., Ltd. 1 Multi Channel Systems GmbH 2 Thomas et al., 1972 Gross et al., 1977 Pine, 1980 Regehr et al., 1989 Nisch et al., 1994 Oka et al., 1999 Litke et al., 2004 Segev et al., 2004 Greschner et al., 2014), but also MEAs with on-chip circuitry limited to the surrounding of the array (Greve et al., 2007) and arrays that include FETs (Offenhäusser et al., 1997) and source follower devices directly wired to circuitry outside the array (DeBusschere and Kovacs, 2001). Multiplexed arrays employ some sort of multiplexing within the actual array (Eversmann et al., 2003a, 2011 Heer et al., 2006 Tokuda et al., 2006 Aziz et al., 2009 Berdondini et al., 2005, 2009 Frey et al., 2010 Huys et al., 2012 Johnson et al., 2012, 2013a,b Maccione et al., 2013 Ballini et al., 2014 Bertotti et al., 2014).

Figure 2. Device comparison. MEA comparison with respect to (A) electrode density and total sensing area, and (B) parallel recording channel count and noise level. (A) For devices with a regular sensor pitch, such as most in vitro MEA devices, the total area is calculated as number of electrodes times the pixel area. For all devices, the number of electrode times the inverse of the electrode density matches the total area. The light gray lines illustrate the number of electrodes. (B) The noise values shown are approximated RMS values stated in the respective citations. The conditions under which these measurements were taken usually differ significantly (such as noise bandwidth, in- or exclusion of electrode noise, inclusion of ADC quantization noise, etc.). Therefore, this graph only serves as a rough comparison. The waveforms to illustrate the noise levels are simulated and have a spectrum typical for MEA recordings. The simulated spikes are typical spikes for acute brain slice measurements recorded with microelectrodes. The recorded amplitudes may vary significantly depending on preparation and sensor characteristics. See Footnotes: 3 , 4 , 5 , 6 , 7 .

Figure 3. Array architectures. This table summarizes and classifies the different architectures that are typically used for MEAs. Advantages, disadvantages are stated and representative selected references given. (A,B) Fixed wiring. (A) Electrodes are directly connected to signal pads with no active circuitry. (B) Electrodes are directly connected to on-chip active circuitry for signal conditioning. (C𠄾) Multiplexed arrays. (C) Signals are multiplexed to the signal pads via column, row addressing in static mode. (D) More flexible addressing is achieved by adding more routing resources within the array in the switch-matrix mode. (E) All electrodes can be sampled at fast speeds in full-frame readout implemented in active pixel sensor (APS) MEAs.

For in vivo MEAs, the connectivity limitation is even more severe, as connections cannot be wired out on all four sides of the array, but only on one of the narrow sides. Figure 2A includes some examples of such devices using fixed wiring (Wise et al., 1970 Najafi and Wise, 1986 Jones et al., 1992 O'Keefe and Recce, 1993 Gray et al., 1995 Bai and Wise, 2001 Csicsvari et al., 2003 Kipke et al., 2003 Blanche et al., 2005 Olsson and Wise, 2005 Fujisawa et al., 2008 Montgomery et al., 2008 Herwik et al., 2009 Du et al., 2011 Berényi et al., 2014) and three recent in vivo MEAs with multiplexing on the shaft itself (Shahrokhi et al., 2010 Seidl et al., 2011 Lopez et al., 2014). For detailed reviews of in vivo MEAs (see Wise et al., 2008, 2004 Ruther et al., 2010).

Figure 2B, on the other hand, focuses only on CMOS-based devices and illustrates the tradeoff between the number of parallel (or quasi parallel) readout channels and the input referred noise of the amplification chain. It illustrates the fundamental fact that a low-noise front-end amplifier requires both area and power. Limiting either will inherently increase the noise levels. The power budget for the entire device, including all circuitry within the array and surrounding it, is limited by the amount of produced heat that one can tolerate. For the area constraints, one has to separately consider the area within the array and surrounding it. Within the array, the electrode density dictates the available area per pixel. Outside the array, the area is limited mostly by the fabrication cost. As a trivial approach to decouple the area requirement from the noise specifications, one can simply place the amplifiers outside the array and directly wire one electrode to one amplifier (Figure 3B). However, this approach still does not allow achieving both a high density and a large electrode count at the same time. Figure 3 lists these fixed-wiring approaches and typical array architectures using multiplexing within the array to overcome this limitation.

Active switching can be integrated into the array, allowing to time multiplex the signals from many electrodes to a few wires that carry the signals out of the array. We now consider two types of time multiplexing, static (Figures 3C,D) and dynamic (Figure 3E) operation (Imfeld et al., 2008). In dynamic mode, each pixel (or electrode) is sampled once within each frame, with typical frame-rates of 2� kHz for CMOS-based MEAs (Eversmann et al., 2003a Johnson et al., 2013b) and some devices allowing as high as 77 kHz (Bertotti et al., 2014). This mode is similar to image sensors used in cameras. Typically, rectangular sub-arrays can be chosen as regions of interest and sampled at faster rates. From a circuit perspective, the challenge in designing full-frame readout MEAs lies in the fact that the multiplexing within the array requires the front-end amplifier to be located within the pixel itself, as the electrode alone exhibits a high impedance and therefore cannot drive the multiplexed readout lines at sufficient speed. Inherently, the available area within the pixels is limited in high-density arrays, making it difficult to build very low noise amplifiers. In addition, the electrodes themselves and the activity within the culture medium show wide band noise (see Section Noise and SNR), thus requiring a low-pass filter within the pixel to prevent noise from being aliased into the signal band due to the sampling. Generally full-frame readout arrays have a high channel count, and therefore the power budget per channel is very limited.

Alternative approaches to circumvent this issue and to allow for devices in which the circuit itself is not the limiting factor with respect to noise performance have been demonstrated. Arrays operating in static mode (Figures 3C,D) have only switches and no amplifiers as active devices within the array. The switches are used to wire electrodes to front-end amplifiers placed outside of the array, where sufficient area for the implementation of low-noise amplifiers is available. This also decouples the number of electrodes from the number of readout channels, which allows budgeting of the available power in more flexible ways. Devices that employ a simple column and row based static addressing are limited in the flexibility of choosing electrodes for parallel readout. A switch-matrix implementation, which consists of a large set of routing wires, routing switches, and local memory, such as SRAM cells within the array, allows the use of complex routing paths to rewire a subset of electrodes to the available readout and stimulation channels in a flexible manner. Often, such an approach is sufficient to observe biological phenomena of interest, as typically not all electrodes exhibit activity. However, experimental protocols tend to get more complex, as one needs to select the “right” electrodes during the experiment. One of the protocols commonly used for such devices is to first scan the entire array in static mode, i.e., record from each rectangular sub block for, e.g., a few minutes, run some online or quasi online data processing on the recorded data, and select a more refined subset based on the recorded activity and the scientific objective of the experiment.

Apart from the array, CMOS devices also require the design of neuronal amplifiers and some sort of data transmitter, either of the amplified analog signals or, more typically, of the already digitized data. Generally, a neural amplifier needs to have high input impedance, which is significantly higher than the electrode impedance, to ensure signal integrity. The amplifier should be of low power to prevent substrate heating that could damage cells or tissue. For in vitro MEA devices, a variety of target applications have to be considered. Therefore, gain and dynamic range requirements can be quite demanding and should be adjustable, such as to cover applications with maximal amplitudes of a few hundred microvolts in acute slice preparations and, on the other hand, up to 10 mV in measurements from cardiomyocytes. The same also holds true for the flexibility in the recording bandwidth. Some applications may require lower frequency signals only, some only spikes in the EAP band, some both bands with different gain requirements at the same time. The circuits need to implement some sort of high-pass filter to block the large 1/f noise of the electrode-liquid interface typically observed. MEA systems can also include stimulation circuitry, covered in the next section, and analog-to-digital conversion (ADC). They need to include an interface to transmit the data and receive commands for controlling the system's operation. The requirements are different for implantable devices, where usually the target application is much more defined, but also the power, reliability, and safety requirements are more stringent. These systems often implement spike detection or classification and wireless transmission in the system, either as a monolithic implementation or hybrid approach using multiple ICs. They may also be powered wirelessly. On the other hand, in vitro MEA systems do not require wireless power or data transmission, as they can generally be directly wired to the data-receiving device. In this case, often common interface standards are employed, such as USB (Multi Channel Systems GmbH 2 ), Ethernet (Frey et al., 2010), National Instrument's DAQ card (Alpha MED Science Co., Ltd. 1 ), CameraLink (Imfeld et al., 2008), or others. Most of these systems support online storage of the full raw data to hard disks, sometimes including some form of lossless data compression (Sedivy et al., 2007).

Many of the circuit requirements can be traded against each other, e.g., one can easily lower the noise by increasing the area or power consumption. The key challenge therefore is to set the target specifications for the given application accurately and optimize the systems for it, without overdesigning specific requirements. Further considerations with respect to noise are given in Section Noise and SNR. Reviews focusing on circuit related issues can be found here: (Wise et al., 2004, 2008 Harrison, 2008 Jochum et al., 2009 Gosselin, 2011).

Stimulation

MEAs allow passive observation, and also active influence and control of neuronal activity. Metal electrodes can deliver electrical stimuli directly using the microelectrodes, whereas for OGFET-based devices, typically an extra capacitive stimulation spot is used to deliver stimuli (Stett et al., 1997). In addition, monolithic CMOS integration of MEAs opens up the possibility to include electrical stimulation circuitry directly on-chip, in turn allowing a high degree of flexibility in generating spatiotemporal patterns of stimulation, higher spatial resolution for stimulation and direct on-chip stimulation artifact blanking or suppression.

Already the very first electrophysiological experiments with frogs by Galvani (1791) involved electrical stimulations using metal wires connected to various sources, e.g., Leyden jars, Franklin's magic squares, and even atmospheric electricity during lightning. In vivo, electrical stimulation is commonly used to stimulate nerves for transmitting sensory information to the brain, such as for cochlear implants (Wilson and Dorman, 2008) and retinal implants (Ahuja et al., 2011 Zrenner et al., 2011) to control, e.g., limbs for neurorehabilitation after nervous system injury and to treat disorders, e.g., Parkinson's disease by deep brain stimulation using brain pacemakers (Montgomery and Gale, 2008). In such applications, the physical distance between the stimulation electrode and target nerves can be rather large, requiring the delivery of high amplitude stimuli.

Lilly et al. (1955) established charged balanced methods using biphasic brief pulses to limit the damage to the tissue and the degradation of the electrodes themselves. Merrill et al. reviewed electrical stimulation using electrodes, listing various materials (Merrill et al., 2005). For in vitro MEAs, effective stimulation protocols were characterized by Wagenaar et al. (2004). The authors studied different stimulation parameters (pulse width, amplitude, pulse shape) that evoke neuronal activity.

One application of electrical stimulation is the use of it as a “trigger,” so-called stimulus-triggered averaging (Cheney and Fetz, 1985). Electrical stimulation allows delivering trigger pulses of high temporal resolution in the order of a few microseconds, depending on the stimulation buffer used and the capacitive load of the electrode. Stimulation can evoke responses with small temporal jitter, e.g., Bakkum et al. observed a jitter of 160 μs using passive MEAs (Bakkum et al., 2008). Bakkum et al. used trigger signals to study the velocity of action potential (AP) propagation in axons of cultured neurons (Bakkum et al., 2013). Figure 4A shows how such stimulus-triggered averages revealed small axonal spikes of different shapes, such as bi- and tri-phasic types. Figure 4B illustrates the reduction in uncorrelated noise with increasing number of averaged repetitions. One potential issue with delivering electrical stimulation to neuronal cells and tissue is the occurrence of artifacts in recording channels, due to the fact that stimulation pulses are typically three to four orders of magnitude larger than the recorded signals. This coupling between stimulation and recording is difficult to prevent, and artifacts are picked up both within the wiring of the array and circuits, but also through the medium of the cell culture or tissue. However, as long as the coupling is purely capacitive, artifacts usually only prevent recording during the stimulation period itself. If the amplitude of an artifact is large, which can occur when a recording electrode is near the stimulation electrode, the artifact may saturate the amplification circuits of the recording electrode. This saturation will prevent recording for an extended period of time after the stimulation ended. Figure 4C shows an example of such a saturated signal from an electrode located 18 μm (center-center) away from the stimulation electrode and a signal without saturation from an electrode located about a 1 mm away. Figure 4D shows the relationship between the distance from stimulation to recording electrode and the duration of saturation for a 11,011-electrode MEA (Frey et al., 2010), without employing any artifact suppression measures. As long as the amplifiers do not fully saturate, it is possible to suppress such artifacts in software by subtracting the estimated artifact (based on templates, filters or local curve fitting) from the data (Hashimoto et al., 2002 Wagenaar and Potter, 2002). To also allow recording from electrodes on which saturation would occur, counter measures in hardware have to be employed. One solution is to use a “reset” switch that can bring back the saturated amplifier into normal operation quickly, by resetting the high-pass filter of the front-end amplifier (Heer et al., 2006 Frey et al., 2010). To suppress artifacts even on the stimulation electrode itself, more sophisticated methods are used. Jimbo et al. proposed a method to decouple the recording amplifiers during stimulation, sample the electrode potential during recording and add the stimulation pulse to the stored electrode potential (Jimbo et al., 2003). This scheme has also been implemented on dedicated ASICs to be used in conjunction with MEA devices (Brown et al., 2008 Hottowy et al., 2012 Tateno and Nishikawa, 2014). Figures 4E,F show stimuli activated neuronal responses with high spatiotemporal precision. In a study to track axonal APs (Bakkum et al., 2013) several ten thousands of stimuli were required, which was possible without damaging the electrodes or cells. In this case, voltage-mode stimulation was used, although the stimulation hardware supported both current- and voltage-mode (Livi et al., 2010).

Figure 4. Stimulation capability of high-resolution CMOS-based MEA. (A) Examples of evoked spikes detected at three sites (columns) along the same axon. The top row shows individual raw traces, and the other rows show traces averaged as indicated. Scale bars, 1 ms horizontal, 10 μV vertical. (B) The amount of averaging necessary to detect a spike with a given height (0.5𠄳 σ) with respect to the detection threshold. (C) Left: A raw voltage trace recorded at an electrode neighboring a stimulation electrode saturated for about 4 ms (flat line). Right: A raw voltage trace recorded at an electrode located 1.46 mm away from a stimulation electrode did not saturate. (D) The duration of a saturated signal occurring after stimuli is plotted vs. distance from the stimulation electrode (mean ± s.e.m. N = 18 stimulation electrodes from five CMOS-based MEAs). Stimuli consisted of biphasic voltage pulses between 100 and 200 ms duration per phase and between ± 400 and 800 mV amplitude. (E) Locations of stimulation electrodes that directly evoked (black boxes) or did not evoke (empty or filled gray boxes) APs detected at a soma located 񾢐 μm away. The line arrow indicates the orthodromic propagation direction. Scale bar, 20 μm. (F) Voltage traces of somatic APs elicited by biphasic voltage stimuli. Traces in response to eight stimuli are overlaid for each of three stimulation magnitudes (indicated at the top), plotted for all effective (black) and four ineffective stimulation sites (gray at the bottom). Stimulation electrode locations are represented as numbered boxes in (E). Scale bar, 200 μV. All panels and description adapted with permission from Bakkum et al. (2013).

Closed-loop experiments, in which neural activity triggers electrical stimulation, employing on-chip stimulation circuitry have been presented by Hafizovic et al. (2007) and Müller et al. (2013). In both cases, the spike detection is performed off-chip on dedicated FPGA hardware. The actual decision to stimulate and the selection of the stimulation waveform patterns is performed on a personal computer in Hafizovic et al. (2007), whereas in Müller et al. (2013) an event engine performing this task is implemented directly on the FPGA platform, making the latency until stimulation shorter and, importantly, reducing its temporal jitter.

CMOS-based devices exclusively devoted to stimulation at high spatio-temporal resolution of close to 7000 electrode per square millimeter and with variable voltage mode pulses have been developed as well (Lei et al., 2008, 2011). Circuit considerations for CMOS-based devices for clinical in vivo application are reviewed (e.g., Ortmanns et al., 2008 Ohta et al., 2009).

Applications of in Vitro CMOS-Based MEAs

In vitro CMOS MEAs have already been used in a wide variety of applications, for recording, for electrical stimulation or for both. Figure 5 lists in vitro CMOS MEAs, their key specifications and preparations for which they have been used so far. Some additional in vitro CMOS-based MEAs that are not listed in Figure 5 can be found here: (Tokuda et al., 2006 Greve et al., 2007 Meyburg et al., 2007 Yegin et al., 2009 Johnson et al., 2012). In addition, the functionality of some in vivo CMOS MEAs has also been demonstrated using in vitro applications (Aziz et al., 2009).

Figure 5. CMOS-based in vitro MEAs. CMOS-based in vitro MEAs, their key specifications and references to biological applications for recording and stimulation are listed in this table. The application list includes only one representative citation for each type of preparation. The specification for each device are taken from the reference listed on top and may differ for other versions of the device.

Certainly, in vitro CMOS-based MEAs, being still an emerging technology with commercial availability only starting recently, have a high potential for future biomedical research and diagnostics (Jones et al., 2011).


4 EEG-Based Motor Imagery Brain Connectivity Analysis

This section reviews previous studies that have investigated the EEG-based brain network of healthy subjects and connectivity analysis during different MI movements. Although great efforts were made to find related studies, only a few have been found, which indicates the short tenure but rapid growth of this research field. In this review, research that highlighted the differences of brain network during both real movement and its imagination are also considered however, only the results for MI movements of different conditions are taken into account and reported here.

In 2004, for the first time, the phase coupling of sensorimotor rhythms in different motor areas during tongue-MI was investigated on three right-handed subjects using PLV (Spiegler et al., 2004). EEGs in channels C3, C4, and Cz were filtered between 1 Hz and 50 Hz and local average reference derivations were computed in order to obtain reference-free EEGs. Then the signals were transformed into the time-frequency domain through Morlet wavelets, PLV measures were calculated between each electrode pair, and the significant PLVs were kept by the use of 95% confidence level. Results reported an increase in amplitude around left and right hand region mu rhythms while imagining tongue protrusion. Phase coupling of 10 Hz oscillations was also increased between the primary sensorimotor hand representation areas (C3 and C4) and the premotor area (Cz). No particular changes in phase coupling were found between mu rhythms in both hemispheres at electrodes C3 and C4.

The usefulness of interaction between EEG signals was examined for classifying mental tasks in the BCI framework (Gysels & Celka, 2004). Using 32 channels, we recorded EEGs from five right-handed volunteers during imagination of repetitive self-paced left hand (LH) and right hand (RH) movements and imagination of words that begin with the same letter. PLV and Coh were measured from sliding 1 second windows, eight times per second within 8 Hz to 30 Hz. The power spectral densities (PSDs) in the (8–12 Hz), (13–18 Hz), (19–30 Hz), and (8–30 Hz) bands were computed by means of the Welch’s averaged PSD. Several subsets of features were made based on the extracted features and fed to support vector machine (SVM) for classification. Pairs of feature subsets were compared with each other to examine significant differences. Generally the results provided evidence of high-variable classification performance within subjects on different days. Classification results indicated that PLV considerably outperformed Coh. and features were better than for two subjects and in the case of all subjects. Low-frequency features performed better than higher-frequency ones. For very short time windows, PLV distinguished different intervals of common activity, though with high inconsistency, while Coh was unable to make any distinctions at all. Moreover, and meaningfully outperformed Coh and PLV. All cognitive tasks were successfully recognized when PLVs and PSD were combined.

Determination of electrical brain activity propagation in beta and gamma band in sensorimotor areas during LH-MI and RH-MI movements was studied on three right-handed participants (Ginter et al., 2005). SDTF measures were estimated where a MVAR model was fitted to 10 EEG signals with model order 5 and 0.4 second window length, which was shifted by 10 samples over the whole epoch. Their results showed a gap in the propagation in alpha and beta bands for electrodes overlying primary motor-sensory area in seconds 3 to 5 (including pre- and perimovement) and bursts of gamma activity in about the same time. A decrease of propagation in beta band was not always accompanied by the increase in gamma sometimes both components fluctuated in a similar way. Investigation on difference of flows in gamma and beta bands after the cue showed that between 3.0 and 3.7 seconds for both hands, the most characteristic feature was the activity in the left hemisphere. For RH, it was the contralateral side, and for LH, where it was the ipsilateral side, the gamma propagation from around C4 was also observed. In epoch 3.7 to 4.2 seconds, for RH, the propagation in gamma band mainly came from primary sensorimotor area (electrode C3) and the locations posterior to it. This activity was mainly directed toward frontal structures. For LH, there was still strong propagation at the ipsilateral side however, the contralateral side became more active, especially in the area underlying C4 and electrodes located closer to the midline. For epoch 4.2 to 5.0 seconds, for RH, the beta-gamma spread from the regions close to C3 and posterior to it (more toward midline) and was directed to the front and to Cz. For LH, this situation was mirrored. For both hands, flows were also observed in more posterior locations at the ipsilateral side.

The pattern of EEG activity propagation in beta and gamma band was also investigated during imagination of lifting up left and right hand index finger (LHF and RHF) movements (Kuś et al., 2005). EEG signals were recorded from nine right-handed volunteers using 30 electrodes. To examine the dynamics of brain activity, signals from 18 channels were analyzed by means of a nonnormalized SDTF based on a MVAR model with model order 5, window size with 400 msec, and sliding window with 80 msec of length. The statistical significance of flow changes related to the subject’s action was then found using a bootstrap technique. The results demonstrated the bigger dynamics of outflows from the primary motor cortex area in comparison to outflows from other areas in beta band. In the same band, the activity was increased after cue onset, and it decreased during MI and a subsequent rebound. For imagination of RHF, 0 to 0.5 seconds after cue onset, the strongest outflow was observed in the left hemisphere. However, this increase was not obvious for LHF in the same period. A short and weak decrease of outflow of beta activity was localized on the right hemisphere around electrode C2, 0.3 to 1.5 seconds after cue onset for LHF. For RHF, the decrease was observed in both hemispheres however, it was more pronounced for electrode Cp3. After the task (1–3 seconds after cue), an increase of outflow was seen mainly at the electrodes at the midcentral and frontal areas. In the gamma band, activity outflow started 1.0 to 1.1 seconds after the cue onset, and big areas of brain, especially around Cz, were engaged in activity emission. In this band, the propagation often started in more posterior areas, which indicated more involvement of the sensory areas. In case of LHF, the gamma activity started to propagate in the right hemisphere (0.4–0.8 seconds after the cue onset), after which the flow was observed with a delay of about 0.7 to 0.8 seconds in the symmetrical locations of the left hemisphere. For the RHF, the results were less consistent in more cases where the flow started in the contralateral hemisphere. Another characteristic feature was the propagation from the central areas of the head (electrodes Cz, Fz, Fc1, Fc2) overlying the SMA, especially for RHF from Fc1 and for LHF from Fc2.

Song and associates (2005) proposed an MI-BCI based on phase synchrony rate (PSR). PSR, computed from binarized PLV, described the number of discrete synchronization events within a window. This study was carried out on data set IVa of BCI competition III (five subjects participated) for RH and foot MI movements. Spatial resolution of raw EEGs was improved by using a Laplacian filter. Then filtered EEGs were split into 6 sliding windows of length 100. Each window was further divided into 76 microwindows (with size 25 and overlap by 24). PLVs were then computed using Hilbert transformation and binarized for each microwindow. Averaging the 76 binarized PLVs resulted in the PSR. As a whole, 6 PSRs were computed for each electrode pair in a trial. PSRs from all electrode pairs were passed to statistical tests and further used as features for classification by linear SVM. Statistical nonparametric tests showed that PSRs contained significant differences between two types of motor imageries. Generally the error of the PSR method was higher than those of the PLV method. The qualitative similarity between the PSR and PLV method suggested that phase was more discriminative than amplitude within the first 1.5 to 2.0 seconds.

An online BCI system based on ongoing phase coupling quantified by the PLV of EEG channels was presented in Brunner, Scherer, Graimann, Supp, and Pfurtscheller (2006). In offline study, EEG signals were recorded from six subjects during imagination of LH, RH, foot, and tongue movement using 22 electrodes. A feature set was formed through a number of PLV features by using signals from seven channels 21 different electrode pairs were constructed. For each pair, PLV features in 12 frequency bands were calculated—namely, 11 nonoverlapping narrow bands (width 2 Hz) between 8 and 30 Hz plus a broadband (8–30 Hz)—yielding 252 different PLV features. These features were fed into a sequential floating forward feature selection (SFFS) algorithm to choose at most eight features and finally served to linear discriminant analysis classifier. It was shown that for most of the subjects, the feature subsets selected by the SFFS contained several PLV values that had been filtered in the broad frequency band between 8 Hz and 30 Hz. Therefore, broad-banded features were chosen to be used in online experiments. Offline topographical analysis indicated that interhemispheric electrode pairs were rarely selected couplings within one hemisphere were dominant in all subjects. Moreover, couplings involving the frontal electrode location occurred more often than the occipital region. Consequently, online study was conducted with only the four pairs and three classes (LH, RH, and foot). The online sessions revealed that all subjects were able to control three mental states with single-trial accuracies between 60% and 66.7% throughout the whole session. Another offline experiment was carried out in order to find out if classification performance can be improved when using PLV features in addition to BP features. Results showed that BP features performed better than phase features when they were combined, the resulting classification accuracy was higher compared to using only one feature type.

The neural coupling level in the mu rhythm during imagination of LH and RH movement was studied using PLV among EEG electrodes in primary motor area (M1) (local scale) and between M1 and SMA (large scale) (Wang, Hong, Gao, & Gao, 2006). EEG signals were captured from six right-handed volunteers by means of 32 channels. Signals instantaneous phase was achieved through the analytic signal using Hilbert transform, and the corresponding phase difference in each signal pair was computed. For large-scale synchrony investigation, PLV was computed between the three pairs FCz-C3, FCz-C4, and C3-C4. For local-scale synchrony, four electrodes around FCz, C3, and C4 were joined to make a five-electrode group, and PLV was computed by averaging over all 10 combinations of electrode pairs from the five electrodes. For single-trial classification, three local-scale and three large-scale PLV features were obtained and fed to a Fisher discriminant analysis (FDA) classifier. Results indicated that large-scale PLV was bigger for RH over C3-FCz, as well as the left M1 area, while the PLV of the LH had a higher value over C4-FCz and within the right M1 area. In addition, PLV indicated a low level of synchrony between left and right hemispheres, and there was no meaningful difference between LH and RH. No significant difference was observed between LH and RH for local-scale synchrony in the SMA area. Classification results showed 84.70% and 77.08% accuracy for large-scale and local-scale synchrony, respectively. A higher accuracy (87.02%) was achieved for the combination of two-scale synchronies, and the best performance (96.13%) was gained when large-scale synchrony and power features were combined.

The previous study was extended by examining amplitude and phase coupling measures for feature extraction in an MI-based BCI (Wei, Wang, Gao, & Gao, 2007). Five right-handed participants were instructed to imagine RH and LH movement, and EEGs were captured with 32 electrodes over the primary sensorimotor area (SM1) and the SMA. The data were re-referenced by common average reference and filtered by spatial Laplacian filter. The connectivity between EEGs was quantified for the amplitude coupling by nonlinear regressive (NLR) coefficient and for the corresponding phase relationship using PLV. The coupling measures were based on five electrodes around C3 and C4 in each hemisphere and three electrodes in SMA: (1) coupling between any two electrodes around C3 and C4 (CW), (2) coupling between each electrode around C3 and each electrode around C4 (CB1), and (3) coupling between each electrode around Fz and each electrode around C3 and C4 (CB2). Six feature vectors were achieved by using these coupling methods separately for computing the NLR and PLV. For single-feature vectors FDA and for a combination of two feature vectors, SVM techniques were applied for classification. In this study, the performance of NLR and PLVs was also compared to AR coefficients, which were estimated from single EEG signals using Burg algorithm. The best classification accuracy was obtained by CB2 for NLR (92.8%) and PLV (92.9%), and CW outperformed CB1 for both coupling measures. Wei et al. (2007) reported that although NLR features delivered slightly higher classification accuracy for CW and CB1, their computational cost was higher than PLV features. It was also shown that considering CW and CB1, AR provided a little higher classification accuracy (93.3%) than the best performance obtained by coupling measures. They finally indicated that a combination of coupling measures with AR coefficients increased classification accuracy.

Stavrinou, Moraru, Cimponeriu, Della Penna, and Bezerianos (2007) investigated the cortical activation and connectivity subserving imaginary rhythmic finger tapping using PS analysis. Three right-handed volunteers were instructed to imagine the kinesthetic of right index finger–tapping movement while EEGs were recorded by 60 channels. Active electrodes (20 electrodes) were selected based on the detection of maximum activity 10 for the left and 10 for the right hemispheres. Laplacian filtering was employed to reduce the effect of VC, and a complex Morlet wavelet transform was used to quantify the oscillatory activity. They applied wavelet PS, taking the scale of the wavelet corresponding to the most reactive frequencies in the beta range. After extraction of phases, the degree of synchronization between any two of the selected electrodes was evaluated on a single trial basis by means of a PS index. Using a significance level of 5%, they reported the frontoparietal coactivation during MI movements and functional connectivity over the contralateral hemisphere. For prestimulus and poststimulus, the most reactive frequencies were in the range of 18 to 20 Hz. A clear decrease in the signal energy was revealed in electrodes C1 and FC3 after the stimulus presentation at the contralateral hemisphere (ERD). Afterward, energy power rebounded, and an ERS occurred. The significant synchronization values identified beta range synchronization between signals recorded at electrodes FCZ, C5 CPZ and CP1.

EEG connectivity during MI of the LH and RH was investigated in a broad frequency range across the whole scalp by combining beamforming with TE (Grosse-Wentrup, 2009). EEG signals were captured from four subjects by 128 electrodes and re-referenced to common average reference. Beamformers were designed to extract those components of the EEG originating in the left and right motor cortex. Then model-based covariance matrices for EEG sources within the left and right motor cortex were computed. The extracted EEG sources, as well as the unfiltered data recorded at each electrode, were then bandpass-filtered with sixth-order Butterworth filters in five frequency bands ranging from 5 Hz to 55 Hz in steps of 10 Hz. Then TE was computed from all EEGs at each sample point. Class-conditional BP changes (ERD/ERS) of extracted sources were computed in order to identify frequency bands with common modulations in BP and TE. Observed changes in TE were statistically significant at level 0.01 for all electrodes. Their results showed no distinct differences in TE between MI of the LH and RH. Instead, the strongest differences in TE were observed in rest versus MI of either hand. The amount of decrease in TE during MI relative to rest increased with higher frequencies and was most pronounced in the gamma band, 45 Hz to 55 Hz. Topographically, the strongest differences were observed in frontal, precentral, and postcentral areas.

Tsiaras, Andreou, and Tollis (2009) investigated the differences between the LH and foot (F) MI synchronization networks by comparing them with the average idle state (I) synchronization network characterized by modern graph theory. EEGs recorded from four participants were filtered by the CSP spatial filter, and PSD was computed by Welch’s algorithm. The synchronization between all channel pairs for mu and beta frequency bands was computed by means of a robust interdependence measure (RIM) and PDC. Results showed that functional networks constructed from motor imagery EEG data were irrelevant to MI movements due to VC effect. Networks LH and F were also very similar. To eliminate redundant connectivities, network I was subtracted from networks LH and F. For two subjects, when networks were constructed by PDC, the network LH minus I had long edges that spanned both hemispheres and was associated with the MI task or occipital or parietal alpha activity. In network F minus I, the long edges were less obvious. When networks were constructed by RIM, the long edges between hemispheres of network LH minus I were mixed with other edges. Network I minus LH and I minus F had more edges close to the LH and F areas of the sensorimotor cortex, respectively. Intersubject variability was also observed in this study.

The interchannel connectivity of LH, RH, feet and tongue MI tasks for a noninvasive BCI application was studied in Chung, Kim, and Kim (2011). The goal was to find a spatiotemporal pattern of connectivity unique to each MI by studying the EEGs of two subjects. They considered a total of 6 seconds of data that contained 2 seconds preimagery and 4 seconds peri-imagery segments. A short-time window approach (1 second window length with step size of 0.1 second) was used to estimate connectivity by computing a linear correlation coefficient (CC) in each window. Their results showed clear differences among the four motor imageries. More positive CCs were obtained for LH and RH compared to feet and tongue. They observed contralateral connectivity for LH and RH in two subjects. CCs between frontal and parietal areas showed substantial difference between LH and RH at different time windows for different subjects. They observed the least number of significant connectivity in the tongue MI in both subjects. They reported the central connectivity for both feet MI in one subject, while no significant channel connection was seen for the other subject.

Hoang, Tran, Nguyen, Huang, and Sharma (2011) proposed a bivariate feature that combined a short-window bivariate autoregressive (CSWBVAR) model for classification of RH and LH motor imagery movements from one subject. Given the channels pair, they divided each of them into overlapping short windows and then estimated bivariate autoregressive (BVAR) parameters for each pair of windows. CSWBVAR was formed by combining extracted BVAR parameters together with a pre-defined overlapping window parameter. Finally, they fed the formed feature vectors into a linear-kernel SVM for classification. Their results showed that CSWBVAR features improved classification accuracy up to 7% compared to the univariate one. This study also investigated the optimum parameters of CSWBVAR feature, window length, and moving step size. Higher accuracy was obtained with a window size of 128 data points and 50% overlapping step size. It was also reported that among the considered channel pairs, C3-C4 produced the most stable and best performance, which emphasized the important role of these two channels in MI problems.

Independent source-based causal connectivity brain network was proposed in Chen, Li, Yang, and Chen (2012) to classify LH and RH motor imagery for BCI applications. At first, scalp EEGs of one subject participated in BCI Competition IV, data set 2a were decomposed via ICA (FastICA algorithm) into maximally independent components. Independent components were then localized using forward and inverse models (equivalent current dipole), while adaptive VAR models were fitted to the time series to model transient information flow in the form of PDC measures. Statistically significant PDCs were obtained using bootstrap for the data set. Then a simple classification rule was defined based on source analysis results and causal connectivity brain network. For equivalent dipole analysis, the classification was correct if the equivalent dipole was located on the contralateral side with the imaginary hand. For causal relation analysis among independent components, the classification was correct if the quantitative feature of causal density and causal flow was related to the corresponding imaginary hand. Based on these two criteria, 86% and 92% classification accuracy was obtained for LH and RH respectively.

Athanasiou, Lithari, Kalogianni, Klados, and Bamidis (2012) inspected the efficiency of the effective connectivity for classifying foot and hand MI tasks, while EEGs were recorded by means of 17 electrodes from seven right-handed participants. EEGs were bandpass-filtered within 8 Hz to 15 Hz, and ICA was employed to get rid of ocular artifacts. Epochs were set from 800 msec prestimulus to 2200 msec poststimulus. A cortical current density (CCD) source model was used to recover the sources of scalp EEGs. These authors compared region of interests (ROIs)—primary hand and foot motor areas (M1), hand and foot sensory areas (S1), and SMAs—activation of foot MI versus hand MI. Effective connectivity was estimated on the whole epoch using DTF derived from the fitted MVAR model. Results indicated that foot and hand MI movements were discriminated for five subjects. There was a stronger activation of SMAs during foot MI for all subjects. Study on the electrode plane revealed that in hand MI, the maximum information flow was from C1 toward FC2 and from FC1 toward C4. On the cortical surface, for hand MI, strong information flow was observed between primary hand motor areas, from the contralateral toward the ipsilateral. In addition, very strong flow was reported from the SMAs toward the ipsilateral primary hand motor area. For foot MI, information flow was from the SMAs toward the contralateral primary foot motor area. Moreover, there was a high bilateral information flow between the SMAs of both hemispheres.

The dynamic of interregional communication within the brain (functional connectivity) during the imagination of LHF and RHF tapping was examined as a control feature for BCI (Daly, Nasuto, & Warwick, 2012). EEG signals were captured via 19 electrodes from 15 subjects with the common average reference montage. Bandpassed-filtered data (0.1–45 Hz) were segmented into trials with 1 second length from cue presentation. The empirical mode decomposition phase-locking spectra was used to map PS levels between all channel pair combinations in each trial. The mean clustering coefficient was then used as a descriptive feature encapsulating information about interchannel connectivity. Hidden Markov models were applied to characterize and classify dynamics of the resulting networks. Their results showed that very high levels of classification accuracy were achieved in the frequency range of 5 Hz to 15 Hz. They reported that the proposed method achieved higher accuracies than the BP approach for all subjects.

Krusienski, McFarland, and Wolpaw (2012) inspected hand MI-based BCI performance when using power spectral, MSC and PLV, separately and in combination. EEGs were recorded through 64 channels, while only 9 channels of the right and left hemisphere hand areas of the motor cortex were chosen for feature extraction. Then fast Fourier transform (FFT), PLV, and MSC were calculated for each trial and pair of channels. Using the extracted features, seven linear regression models were constructed for classification: PLV, MSC, FFT, PLV + MSC, PLV + FFT, MSC + FFT, and PLV + MSC + FFT. The model weights were set by means of a stepwise linear discriminant analysis. Results showed that PLV performed better than only PLV + MSC, which was outperformed by FFT, PLV + FFT, and PLV + MSC + FFT. It was indicated that the FFT-based feature was at least as effective as the PLV and MSC features. Furthermore, inclusion of PLV and/or MSC in models with the FFT based did not improve the performance compared to the FFT based alone. It was concluded that PLV and MSC-based features did not offer more information than FFT-based features.

In 2013, the brain effective connectivity network was investigated to understand brain function and compare the network between RH-MI movement and rest state (Li, Ong, Pan, & Ang, 2013). EEG signals were recorded by 27 channels from eight subjects during MI of RH and rest state (mental counting). EEGs were bandpass-filtered 8 Hz to 35 Hz, and for each trial, time segments of 0.5 to 2.5 seconds after the cue were used for analysis. The MVAR model was built up based on raw EEGs then PDC and DTF were calculated and integrated in the range of alpha band. A one-sided t-test was applied to compare the averaged PDC and DTF of each pair of connections between MI trials and rest trials. Results showed that one subject had very high information flow from the central motor cortex. Another subject showed high activation in the left motor cortex, although the significant sources were not concentrated around C3. They also observed the prominent source at FC4, sensorimotor area. For another subject, strong source activation was observed in the posterior left motor cortex. Cross-validation accuracy of around 70% was reported for the subjects. For the rest of the subjects, they noted that less significant influential sources were located in the left motor cortex.

Later, Billinger et al. (2013) extracted single-trial connectivity measures from VAR models of independent components for classification in a BCI setting. In this study, 45 channels were used to record the EEGs from 14 volunteers while performing hand and foot MI movements. At first, an extended Infomax ICA was used for extracting the source signals. Then, connectivity estimates were measured from all sources with a window length of 4.5 seconds. For further process, eight sources were chosen after being ranked based on the estimates. The final connectivity estimates were computed for these sources with window length 1.5 seconds. BP was calculated by FFT, and connectivity measures cross-spectral density, Coh, PDC, PDCF, GPDC, DTF, ffDTF, direct DTF (dDTF), and directed coherence were computed from VAR coefficients. Modified false discovery rate was then used to find statistically significant connectivities, and shrinkage linear discriminant analysis was employed for classification. Results showed that ffDTF, dDTF, and BP performed similarly while outperforming other measures. They reported that Coh and the unmodified DTF were not appropriate for BCI.

Gonuguntla et al. (2013) analyzed the network mechanisms related to LH and RH motor imagery tasks based on PLV in EEG alpha band. This experiment was performed on the eighth subject of BCI Competition IV data set 2a. They computed the difference of PLV between active and rest states for all electrode pairs in the range of 6 Hz to 14 Hz. The five most significant pairs (MSP) corresponding to the pairs with maximum difference PLV were selected for each task. MSPs were laid in the contralateral part corresponding to task—5 MSPs for the left side and 5 MSPs for the right side of the brain. MSPs were employed as features, and the difference in PLV level after cue was used for classification. Results showed that the PLV magnitude increased during the imagination compared to rest state within the frequency band 9 Hz to 11 Hz. It was demonstrated that the (Cz,Cp3) pair was the most significant one for RH-MI, which was mostly seen during imagination. Similarly the (Fz,Cp4) pair was identified as MSP for LH imagery. Classification results indicated the potential of such methodology for BCI applications.

In 2014, Hu, Wang, Zhang, Kong, and Cao (2014) used scalp EEGs at Cz, C3 and C4 from data set 2b of BCI Competition IV to study causality flow during MI. They proposed a new causality (NC) in time and frequency domains using a time-invariant BVAR model. They found strong directional connectivity from Cz to C3/C4 during LH and RH motor imagery. During LH-MI, there was directional connectivity from C4 to C3, whereas during RH-MI, there was strong directional connectivity from C3 to C4, which was more clearly revealed by NC than GC. They concluded that NC in time and frequency domains was much better than GC to reveal causal influence between different brain regions.


Chapter 7

Examiner taps a sequence of blocks (i.e. 4, 5, 1, 8, 2). The Participant then has to repeat the sequence. Block-span.

Difficult because movements are backward in the mirror. With practice, participants figure it out, and show recall when retested days later.

Same place each trial day. for the whole day. But switches day to day.

1. These studies tell about the injured region's functions as well as what the remaining brain can do to in absence of the injured region.

Device/surgical instrument that permits a researcher/neurosurgeon to target a specific part of the brain for ablation.

To avoid compensation following permanent lesions, a hollow metal coil is placed next to neural structure then chilled fluid is passed through, cooling the brain structure to about 18 degrees C.

When chilled fluid is removed from coil, the brain structure quickly warms, and synaptic transmission is restored.

Noninvasive technique. Takes advantage or relation between magnetism and electricity

Based on the discovery that light can activate certain proteins that occur naturally and have been inserted into cells of model organisms.


EEG Signal Analysis: A Survey

The EEG (Electroencephalogram) signal indicates the electrical activity of the brain. They are highly random in nature and may contain useful information about the brain state. However, it is very difficult to get useful information from these signals directly in the time domain just by observing them. They are basically non-linear and nonstationary in nature. Hence, important features can be extracted for the diagnosis of different diseases using advanced signal processing techniques. In this paper the effect of different events on the EEG signal, and different signal processing methods used to extract the hidden information from the signal are discussed in detail. Linear, Frequency domain, time - frequency and non-linear techniques like correlation dimension (CD), largest Lyapunov exponent (LLE), Hurst exponent (H), different entropies, fractal dimension(FD), Higher Order Spectra (HOS), phase space plots and recurrence plots are discussed in detail using a typical normal EEG signal.

This is a preview of subscription content, access via your institution.


INTRODUCTION

Methodological advances in neuroscience research have enabled novel approaches to investigating how the brain supports dynamic real-world social interactions. For example, researchers have begun to study the neural basis of social interactions by comparing the brain responses of multiple individuals during a variety of seminaturalistic tasks (for a review, see Hasson & Frith, 2016 Babiloni & Astolfi, 2014 Scholkmann, Holper, Wolf, & Wolf, 2013 Hasson, Ghazanfar, Galantucci, Garrod, & Keysers, 2012). Research involving turn-taking in gestural (Schippers, Roebroeck, Renken, Nanetti, & Keysers, 2010) as well as verbal (Dikker, Silbert, Hasson, & Zevin, 2014 Stephens, Silbert, & Hasson, 2010) communication have demonstrated a relationship between brain synchrony and comprehension as well as the predictability of another person's communicative act. Further work has shown that complex audiovisual stimuli (e.g., natural movies) elicit similar brain activity among viewers and emotional responses and, crucially, vary as a function of participants' attentional engagement (Ki, Kelly, & Parra, 2016 Chang et al., 2015 Nummenmaa et al., 2012 Jääskeläinen et al., 2008 Hasson, Nir, Levy, Fuhrmann, & Malach, 2004).

Although these experiments explore the similarities and differences in neural activity across participants as they engage in similar or pseudointeractive tasks, they do not capture the dynamic nature of real-world settings. Methodological constraints limit the ways in which researchers have been able to explore the brain basis of social interactions as they occur in real world. Although providing promising results, these studies are still largely confined to the laboratory, mostly limited to dyads, and typically use neuroimaging technology with low temporal resolution (e.g., functional near-infrared spectroscopy). We know that the direct study of face-to-face exchanges is critical to fully understand social interactions, yet there is a gap in the research exploring the underlying neural mechanisms of joint behavior as it naturally unfolds (Dumas, 2011). To be able to investigate how the brain supports interactions that resemble the complexity of the interactions we encounter in everyday life, hyperscanning research will have to accommodate more ecologically valid situations (Babiloni & Astolfi, 2014 Schilbach et al., 2013 Dumas, 2011). In the current study, we investigated the neuroscience of real-world classroom learning using mobile electroencephalography (EEG) headsets to simultaneously record participants in support of previous experimentation by Dikker et al. (2017).

Increasingly, research shows that, during joint actions, people become “coupled” at motor, perceptual, and cognitive levels in both planned and improvised coordination (Knoblich, Butterfill, & Sebanz, 2011). Participants during synchronized motor activity modify their own actions in response to their partners (Dumas, Nadel, Soussignan, Martinerie, & Garnero, 2010). Hyperscanning neuroscience research has shown not only a relationship between synchrony at the motoric and neural levels (Dumas et al., 2010) but also that face-to-face interactions moderate the relationship between social factors and brain-to-brain synchrony (Dikker et al., 2017 Jiang et al., 2015 Hari, Himberg, Nummenmaa, Hämäläinen, & Parkkonen, 2013 Scholkmann et al., 2013 Jiang, Dai, Peng, Liu, & Lu, 2012 Dumas et al., 2010). Specifically, joint action tasks demonstrate that synchronous motor activity within interactive partners leads to increased feelings of affiliation and social cohesion (Valdesolo, Ouyang, & DeSteno, 2010 Hove & Risen, 2009 Bernieri, 1988), particularly in cooperative versus competitive contexts, and that this is reflected at the neural level (Cheng, Li, & Hu, 2015 Cui, Bryant, & Reiss, 2012 Yun, Watanabe, & Shimojo, 2012).

The classroom setting is an exemplary environment to systematically investigate group interactions—between students and students with their teacher—under semicontrolled conditions, while measuring behavioral and cognitive outcomes (e.g., academic performance and student engagement Scholkmann et al., 2013 Watanabe, 2013). The dynamic interaction between a teacher and a group of students is fundamental to classroom learning and has been shown to affect both student engagement and academic achievement (Watanabe, 2013 Hughes, Wu, Kwok, Villarreal, & Johnson, 2012 Walton & Cohen, 2011 Hamre & Pianta, 2001 Bernieri, 1988). Teaching and learning can be viewed as a joint action between the teacher and the students such that features of the interactive partner and the event are treated as stimuli in a reciprocal exchange (Sensevy, Gruson, & Forest, 2015). Research into student–teacher relationship exchanges in the classroom suggests that exploring underlying neural activity may support understanding and predicting educational outcomes from the perspective of the teacher and the student (Holper et al., 2013). Recently, researchers have used portable EEG equipment in the classroom to record nine students simultaneously during natural movie viewing and reproduced findings from similar, laboratory-based, experimental designs with commercial-grade equipment, demonstrating the potential for real-world measurement of students' attentional engagement (Poulsen, Kamronn, Dmochowski, Parra, & Hansen, 2017).

In further recent classroom-based experimentation, which forms the foundation for the current work, authors report that brain-to-brain synchrony (quantified as total interdependence [TI] or interbrain coherence Wen, Mo, & Ding, 2012) between students during class activities was correlated with student engagement and classroom social dynamics (Dikker et al., 2017). Students' synchrony to the group was higher in their preferred teaching style (e.g., video over lecture) and related to greater student focus, group affinity, and empathy (Dikker et al., 2017). In addition, findings in group social dynamics speak directly to the presence of others as a moderator of student synchrony during class. For example, higher student ratings of their teacher correlated with a smaller difference between video (where the teacher played no role) and lecture conditions (where the teacher was central), and students who engaged in prelesson face-to-face baseline recordings showed the highest pairwise synchrony during class with their mutual gaze partner compared with other random students in the group (Dikker et al., 2017). Together, their results suggest that brain-to-brain synchrony is driven by a combination of (i) stimulus properties, (ii) individual differences, and (iii) social dynamics.

The Current Study

In the context of classroom learning, attention is known to play a critical role in learning and maintaining information (Reyes, Brackett, Rivers, White, & Salovey, 2012), and student attention is a challenge even for the most experienced teachers (Evertson & Weinstein, 2013). If brain-to-brain synchrony indeed increases as a function of shared attention (to the teacher, the lesson content, peers), as suggested by the research summarized above (Dikker et al., 2017 Ki et al., 2016), and attention increases retention (Cohen & Parra, 2016), we can then ask whether a student's neural synchrony to the rest of the group or with the teacher predict their retention of the content.

Does brain-to-brain synchrony between a student and their peers predict their retention of the class content?

Is there a relationship between student–teacher brain-to-brain synchrony, classroom learning, and student–teacher relationships, respectively?

In Dikker et al. (2017), both student ratings (e.g., engagement) and brain-to-brain synchrony between students were higher when students viewed lesson-related videos compared with lectures, which allows us to ask if such a parametric difference also exists for content retention (Research Question 1). In addition, as the teacher plays a pivotal role during lectures but not during videos, we ask whether the student–teacher relationship matters more when the teacher is present (Research Question 2). To address these questions, we employed a similar setup as the classroom EEG findings from Dikker et al. but included two metrics in addition to student-to-group synchrony: (1) student performance and (2) student–teacher brain-to-brain synchrony.


Difference Between EEG and MRI

Nowadays, most disease conditions have been widely studied and researched in order to develop the most practical and surest way for cure and relief. Throughout the years, diseases have plagued millions of people around the world. Doctors and researchers have continuously stepped up their efforts to develop new methods and procedures to make sure that they are treating the right conditions. With so many diseases that may have similar signs and symptoms, it is up to sophisticated machines and diagnostic procedures to ensure that they are able to find the source of the problem.

With the huge improvements in disease diagnosis and testing, it is undeniable that we read breakthroughs in curing certain conditions that were once thought to be incurable and deadly. Furthermore, physicians have soon developed newer and safer methods of testing their patients in order to prevent further or additional harm, as well as, incorrect diagnosis of disease conditions. With the life of a patient at stake, it is therefore imperative that physicians use the best diagnostic procedures that can help them make sound and accurate decisions.

There are a lot of diagnostic procedures using diagnostic tools that have different functions and uses. These machines have revolutionized the way physicians work, and it has made their judgments and diagnosis more accurate. Among these machines, an EEG and a MRI have been noted to effectively help make accurate findings about the condition of the body. However, they are entirely different from each other.

First, an EEG is the acronym for an electroencephalography. It is a diagnostic test using a special machine that detects brain wave activity and functioning. The machine is attached to the scalp to record the electrical impulses generated by our brain. Basically, our neurons fire off electrical stimuli which is detected and recorded by this machine. It is then read or analyzed by expert physicians who would look for electrical abnormalities in the results or findings. Depending on a suspected condition, physicians would look for abnormal brain wave activities, for example, spikes or sharp waves which are usually noted in children with epilepsy. This is basically how an EEG is conducted.

On the other hand, a MRI stands for Magnetic Resonance Imaging. It is considered as an advanced diagnostic procedure that uses magnets and radio waves to picture a part of the body being tested. Furthermore, it is a non-invasive procedure that will visualize any internal part of the body. The concept relies on a magnetic field passed into our body, which then creates an image of the body under study. With that, any abnormalities and anomalies are detected and seen.

If you want to know more, you can read further since only basic details are provided here.

1. Diagnostic procedures are reliable ways to determine what is wrong with the body.

2. EEG analyzes brain wave functioning using electrical impulses generated by the neurons.

3. MRI focuses a magnetic field into the body to create an image and look for any anomalies.


Abstract

Background and Purpose—Multimodal neuroimaging with positron emission tomography (PET) scanning or functional MRI can detect and display functional reorganization of the brain’s motor control in poststroke hemiplegia. We undertook a study to determine whether the new modality of 128-electrode high-resolution EEG, coregistered with MRI, could detect changes in cortical motor control in patients after hemiplegic stroke.

Methods—We recorded movement-related cortical potentials with left and right finger movements in 10 patients with varying degrees of recovery after hemiplegic stroke. All patients were male, and time since stroke varied from 6 to 144 months. All patients were right-handed. There was also a comparison group of 20 normal control subjects.

Results—Five of 8 patients with left hemiparesis had evidence of ipsilateral motor control of finger movements. There were only 2 cases of right hemiparesis in addition, 1 patient had a posteriorly displaced motor potential originating behind a large left frontal infarct (rim).

Conclusions—Reorganization of motor control takes place after stroke and may involve the ipsilateral or contralateral cortex, depending on the site and size of the brain lesion and theoretically, the somatotopic organization of the residual pyramidal tracts. Our results are in good agreement with PET and functional MRI studies in the current literature. High-resolution EEG coregistered with MRI is a noninvasive imaging technique capable of displaying cortical motor reorganization.

Recovery that continues beyond 3 or 4 weeks after a stroke has been attributed to neuroplasticity, a reorganization of the brain in which functions previously performed by the ischemic area appear to be assumed by other ipsilateral or contralateral brain areas. Neuroplasticity has been variously attributed to redundancy (parallel distributed pathways), changes in synaptic strength, axonal sprouting with formation of new synapses, assumption of function by contralateral homologous cortex, and substitution of uncrossed pathways. 1 2 Transcranial magnetic stimulation, 3 positron emission tomography (PET), 4 and functional MRI (fMRI) 5 have been successfully applied to demonstrate cortical reorganization after hemiplegia. We applied the new modality of 128-electrode high-resolution electroencephalography (hr-EEG) 6 to detect reorganization of cortical motor control after spinal cord injury. 7 8 We now report that hr-EEG recording of movement-related cortical potentials (MRCPs) can also reveal cortical motor reorganization after recovery from hemiplegic stroke. We evaluated the motor potential (MP) component of the MRCPs in the present study. Direct subdural recordings in humans have shown that the MP has maximal amplitude over the contralateral somatosensory cortex. 9 The MP has been interpreted as the cortical activation of the final common pathway, the pyramidal tract. 10 The recording of the MP may provide a useful tool for the localization of cortical motor control.

Subjects and Methods

We studied 10 outpatients and 20 control subjects. All were male, and the average age was 59.4 years (range 50 to 71 years) (see Table ). None were acutely ill. Eight had a right hemiparesis and 2 a left hemiparesis. Average time after stroke when studied was 43.4 months (range 6 to 144 months). Seven patients had diabetes mellitus, and 3 were hypertensive. One patient had a capsular lesion, 2 had infarcts confined to the right or left cerebral cortex, 1 had bilateral brain stem and cortical lesions, 1 had frontal and pontine infarcts, and 5 had involvement of cortex and basal ganglia. All patients received rehabilitation therapy, and all showed some degree of recovery (Table ). None were noted to have mirror movements. The Institutional Human Use Committees approved the study, and all subjects gave informed consent.

Data Acquisition and Analysis

MRCPs were recorded in the EEG with movements of the index or middle fingers, and the MP component was selected for mapping and dipole source localization studies. There is agreement that MRCPs 11 consist of (1) the Bereitschaft potential, with onset 1 to 1.5 seconds before self-paced movement, followed by (2) a steep negative slope ≈500 ms before movement, then (3) a brief premotor positivity at 50 ms before movement onset, and finally (4) a sharply rising MP, which may begin shortly before movement. MP latencies are measured from movement onset to peak negativity. The MP peak is the highest negativity reached in the scalp-recording motor cortex. The MP component was selected for mapping and dipole source localization studies because it is recorded over the hemisphere contralateral to the finger movement and generated mainly in the primary motor cortex (M1). 11 12

An electrode cap made from stretchable fabric and containing 120 scalp electrodes encased in plastic holders was used. The cap was put on the head with reference to the landmarks of the nasion, inion, and preauricular notches and was stretched to properly position the electrodes. There was an estimated average interelectrode distance of 2.25 cm. Two other channels were used to monitor horizontal and vertical eye movements, and 1 channel was used for electromyogram (EMG) recording. Individual scalp sites were slightly abraded through the hole in the top of each electrode, and conducting gel was injected. Electrode impedances were lowered to <5000 Ω.

An electromagnetic digitizer (Polhemus) was used to sample the surface of the head and the electrode positions on the scalp to establish the accurate location of electrode coordinates in 3-dimensional space. Five thousand to 7000 points were obtained and entered into the host computer as an individual file, which was interfaced with MRI. The 128-channel DC amplifier system (Neuroscan) was calibrated. Data acquisition was set at a digitization rate of 500 Hz for continuous recording. Filter band pass was from DC recording to 100 Hz. At a gain of 1000, the dynamic range was 5 mV, with resolution of 0.084 μV/bit. Scalp electrodes were referred to the ipsilateral ear during data acquisition and for comparisons were rereferenced to an average reference. The Neuroscan system digitized 128 channels simultaneously and displayed topographical maps. Each epoch of EEG recorded at 120 electrodes was individually scrutinized for artifact and either included in the average or rejected.

Subjects were seated on a reclining chair or in a wheelchair or were placed prone on a bed. The subjects were asked to rapidly flex and extend the middle finger (or index finger if movement of the middle finger was not possible) every 7 to 10 seconds. The averager was triggered by the rectified EMG signal recorded by bipolar surface electrodes placed over the appropriate muscles in the forearm. Recordings were made of fingers on each side sequentially. For each digit tested, 3 blocks of 70 movements were recorded for offline averaging. The EEG was averaged for 2 seconds before and 1 second after the EMG onset. The MP was averaged and graphed for all 120-electrode locations and displayed in electrical field maps.

Dipole Source Analysis

Dipole source analysis was accomplished with a current reconstruction and imaging software package known as CURRY Multi-modal Neuroimaging. 13 This package used several reconstruction algorithms (eg, single dipole, multiple dipole, and current density distribution) with subject-specific MR images to restrict the volume-conductor geometry to the individual anatomy. The shape of the optimum volume-conductor model was determined by segmentation or separation of the skin, skull, and brain surfaces from the MR image. We modeled these compartments using the boundary element method by assigning different appropriate conductivity values for each surface (eg, skin and skull). This allowed accurate localization of cortical activity by restricting the model to neurophysiologically appropriate source locations, such as the cortex. Calculations were based on a window of 50 ms before and after the MP peak for dipole analysis. We limited our use of dipole source analysis to the comparison of the MP distribution fields with their putative sources in individual subjects. The spatial localization of dipole sources of MRCPs has been shown to be accurate, and with self-paced movements, a single dipole can be found with low variance, ie, 5% to 10%.

Results

Cases 1 through 7 and case 9 had infarctions of the right hemisphere and left hemiparesis. Case 8 had right frontal, pontine, and cerebellar infarctions causing right hemiparesis and ataxia. Case 10 had left frontal and basal ganglia infarctions with right hemiparesis and aphasia. MPs were recorded in all patients. (See Table .)

Right Hemisphere Infarction

In cases 5 and 9, each with left hemiparesis, the MPs were recorded in a left frontal location with movements of either left or right fingers (contralateral to the lesioned hemisphere). The cortical infarct was right frontal in case 5 and right temporoparietal in case 9. There was basal ganglia involvement in both cases. Figures 1A and 1B (case 5) display the MPs averaged at each of 120 electrodes with movements of the right and left index fingers. With right finger movement, there was a normal distribution of the MPs, with higher amplitude on the left. Movements of the left finger were also associated with higher amplitude on the left, a paradoxical result. The grand averages of all the electrodes produced the summations shown in Figure 2 , where left-sided MPs were associated with both left and right finger movements. This result was not present in any of the normal control subjects. Figure 3 shows the MP averages in case 9, with the addition of dipole source localization. The dipole generators of both left and right finger movements originated in the left hemisphere. An extensive infarction can be seen on the right (in this Figure , the image of the right hemisphere is on the viewer’s left).

In cases 2 and 3, movements of the affected left fingers were associated with centrally placed MPs, whereas the MPs were contralateral (left frontal) with right finger movements. Figure 4 (case 2) combines MPs, current densities, MRI, and source localization. With left and right finger movements, the current densities were more intense over the left hemisphere. The dipole generators originated in the left hemisphere. In case 6, the patient had a recent white-matter infarct adjacent to the left lateral ventricle and a prior occlusion of the left posterior inferior cerebellar artery. He had a left hemiparesis. Bilateral calcification of the carotid and vertebral arteries was present. There was a central MP with left finger movement and a left-sided MP with right finger movement (not illustrated). The latency of 535 ms with left finger movement was exceptionally long.

Left Hemisphere Infarction

In case 10 (Figure 5 ), the MP with left finger movement was in a normal location. In contrast, the MP with right finger movement was in a posterior position. There was an extensive infarction of the left hemisphere (Figure 5 ), with encephalomalacia involving the frontal lobe and extending into the basal ganglia. Figure 6 shows the dipole source originated from behind the infarct (the rim) in the involved hemisphere. The patient had a nonfluent aphasia with a right hemiparesis.

Cases 1, 4, 7, 8, and 10 had contralateral dominant MPs with movements of the affected side. The mean latencies in the contralateral group (n=5) were longer than in the ipsilateral group (cases 2, 3, 5, 6, and 9), but only with movements of the right finger. Otherwise, there were no differences with respect to lesion location, recovery, age, or duration of stroke. Latencies were not different between intact and affected side movements, but the numbers were small.

In summary, normal individuals had MP field distributions at 120 electrodes that, when averaged, localized to the hemisphere contralateral to the finger movements. Patients recovering or recovered from right hemisphere infarcts had MPs, which mapped either to the left hemisphere or a central location with finger movements of the left hand. A patient with right hemiparesis and aphasia had a dipole source at the posterior rim of a large frontal infarct when moving his right finger.

Discussion

We have demonstrated that hr-EEG coregistered with MRI is capable of identifying reorganization of motor control in stroke patients with recovered hemiplegia. This may involve a total or partial shift to the ipsilateral hemisphere, or there may be relocation in the same hemisphere. Honda et al 14 recorded MRCPs, regional cerebral blood flow measurements, and PET activation studies in 2 patients with hemiparesis. Their multimodal studies suggested an important role of the ipsilateral hemisphere in the process of motor recovery. Recently, Cao et al 15 reported in an fMRI study that sensorimotor cortex in the intact hemisphere was activated in 6 of 8 patients during movements of the affected hand. In addition to observing such ipsilateral activation, we noted a contralateral posterior displacement of the MP to the rim of the peri-infarct area in our case 10. Although the patient had a left internal carotid artery occlusion with right hemiparesis, the right sensorimotor cortex was not activated, perhaps because the right internal carotid artery was stenosed. There has been much speculation about the identity and role of compensatory processes in stroke recovery. Cramer et al 16 in their fMRI study of patients recovering from stroke suggested 3 processes related to motor recovery, namely, “activation of a motor network in the unaffected hemisphere greater than seen in controls, increased degree of SMA [supplementary motor area] activation, and activation of foci along the rim of a cortical infarct.” Our methodology did not allow us to record the degree of supplementary motor area activation, but we can confirm their other 2 conclusions. Despite completely different methodologies, our EEG studies have also shown both activation of a motor network in the unaffected hemisphere much greater than that seen in controls and activation of foci along the rim of a cortical infarct. The dipole source in our case 10, originating behind the rim of a left frontal infarct, suggests a reorganization occurring in the area of the former ischemic penumbra. 17 There was no evidence of ipsilateral activation, probably because of right hemisphere ischemia due to an internal carotid occlusion.

Further understanding of how manifestations of neuroplasticity relate to stroke recovery in this and other cases should come from the continued application of multimodal techniques in brain imaging. 18 The combination of fMRI and hr-EEG may be complementary, because 1 modality (fMRI) has greater spatial and the other (hr-EEG) better temporal resolution.

Figure 1. A, Case 5: right finger movement. Left central and parietal electrodes are recording the MP negativity. Patient had a left hemiparesis. B, Case 5: left finger movement. Left central and parietal electrodes are recording the MP negativity. An MP is recorded at some right central electrodes, but amplitude is lower than on the left. Patient had a left hemiparesis.

Figure 2. Cases 9 (A) and 5 (B). Both patients had left hemiparesis maps of the averaged MPs with left (LF) and right (RF) finger movement are shown for both patients. MPs are the left-of-center areas of negativity (dark end of the gray scale) in the maps.

Figure 3. Case 9. MPs with either left or right finger movement are on the left side of the maps and were best recorded in the left hemisphere. Bottom, The dipole sources (arrows) within the left hemisphere were oriented inferiorly and toward the right. The beginning of the dipole arrow identifies the source, the size demonstrates the strength, and the point identifies the direction. Note the large right temporal lobe lesion in the MRI.

Figure 4. Case 2. Juxtaposition of MPs, current densities, and dipole sources. The MP with left finger movement is in a central rather than a normal right (contralateral) position. The current density field shows activity bilaterally (yellow, green, and red, with yellow indicating peak density, red the least density, and blue, no density). The dipole source originates in the left hemisphere, close to the midline (red arrows). Right finger movement is associated with a left (contralateral) MP, as expected. The electrical field differs from the right hemisphere field in that it is more active (yellow and green) and localized close to the left sensorimotor cortex. The dipole source originates in the left hemisphere. Patient had an infarction of the right internal capsule and a left hemiparesis.

Figure 5. Case 10. The MP with left finger (LF) movement is in a normal location. The MP with right finger (RF) movement, in contrast, is in a posterior position. MRI reveals a large infarct on the left with dilated ventricle. Patient had right hemiparesis and aphasia.

Figure 6. Case 10. 1, General cortical atrophy with left-sided infarct. Note that the dipole with right finger movement originates in the left hemisphere, behind the infarct (“rim”). This is well seen also in the sagittal view (2), in which the dipole originates just behind the infarct.

Table 1. MP Locations With Right or Left Finger Movement

RF indicates right finger LF, left finger Post, posterior location of MP C, central location of MP and LAT, latency of MP.

1 Rehabilitation results are shown on a scale of 0 to +++++.

This study was supported by the Office of Research and Development and Rehabilitation Research and Development Service, Department of Veterans Affairs, Washington, DC. The authors wish to express their gratitude to the patients who volunteered to participate in this research and to their family members who accompanied them. Thanks are also due to the medical staff of Hines Veterans Affairs Hospital for their referrals.