Essay details:

  • Subject area(s): Engineering
  • Price: Free download
  • Published on: 7th September 2019
  • File format: Text
  • Number of pages: 2

Text preview of this essay:

This page is a preview - download the full version of this essay above.

Human beings differentiate themselves from other species by their ability to express and share thoughts, emotion with others. The brain is the central unit of this communication. From an engineering point of view, human brain is characterised by every feature and property that engineers would love to avoid. They are incredibly complex, and their processes are chaotic, unstable, non-stationary, noisy and unpredictable [5].Thus accessing and processing brain signals  commonly referred to as EEG to predicts one’s cognitive state and translate thoughts into action have been of a growing interest among the science community over the last couple of years leading to the emergence of BCI technologies [5].

BCI are devices that enable people to interact with their surrounding via employing bioelectrical signal generated from their brain activity. The bioelectrical signal which is often in the order of microvolts is measured by placing a symmetrical pair of electrodes on the user scalp [1]. Based on the way by which these brain signal are being measured, BCI system can be categorised as invasive type where the electrodes capturing the EEG signal are directly implanted on the user’s cortex via risky surgical operation or non-invasive type system which are safer but provides a poor quality signal since the electrodes a placed on the user’s scalp [2].

Thus, the efficiency of a BCI system in translating user’s intents into action highly relied on the process and instrument used to acquire the EEG signal arising the from the brain when performing mental tasks [6]; most commonly there are high level and expensive EEGs recording system mainly used for medical application (e.g. INFANT: Irish Centre for Fetal and Neonatal Translational Research) that provide good a signal resolution hence easing the analysis of the brain activity. On the other hand, low-level headset such as the Emotiv Epoc SDK, which is a consumer-priced EEG recording devices mainly design for the gaming industry.

However, whether the BCI system intended usage is to provide people suffering from severe motor impairments a muscular-free communication channel with their social environment along with restoring a sense of autonomy to these patients or use to assist the doctor in monitoring the cognitive state of patient in prolonged coma [7], all BCI system depend on the techniques and effectiveness of the methods used to extract discriminative features from the recorded EEGs signals [6]. In order to provide a better understanding of the brain activity and ease their processing. The electroencephalographic signal (EEGs) are often classified in different types brain waves signals based on the frequencies bands at which they occurs [6]: Delta waves [0.5-4 Hz] associated with profound sleep; Theta Wave [4-8 Hz] appears during emotional stress; Alpha Wave [8-14 Hz] observed when closing the eyes and relaxations state. And finally Beta Wave [14-30 Hz] more present during thinking activities.

In light of the above, most brain-computer interface application analysis and interpret these signals to obtain feedback from the user cognitive state. For instance [8] used one of the Alpha brain waves oscillations properties which is the increase in their amplitude observed during wakeful relaxation with eyes closed to implement an alertness estimation BCI system. The level of attention of student and car drivers were monitored by processing EEG signal generated from their brain as a result of a light stimulus. More precisely, the Alpha wave signal were extracted via wavelet transform and the change in amplitude of the signal along with the entropy energy of the WT coefficients were examined to determine the level of attention of the test subject; the paper [8] claims to have obtained 81% success rate. Similarly, abrupt changes observed in EEGs signal can also be used to implement other non-invasive human machine assistive care technologies such as a P300 speller, where an amyotrophic lateral sclerosis patient can select words from a virtual keyboard. In [9], event-related potentials (ERPs) elicited during decision taking action known as P300 signals were used to develop BCI based P300 speller. P300 signals are characterised/ detected by a positive rise in amplitude (peak) of the EEGs around (250-400 ms) after an applied mental stress. The paper [9] uses the famous Row-Column paradigm to design the speller system; the patient was instructed to mentally select a letter on virtual keyboard made of 6×6 matrix of characters randomly highlighted at a certain interval, once a target letter is flashed, a P300 signal is detected and correlated to the corresponding and displayed on the screen [9]; the process is then reiterated until when the patient desired phrase or expression is achieved.

Clearly, significant ground-breaking works have already been made in relation to BCI technologies. Furthermore, the development of new technologies coupled with the advancement in applied neurology and signal processing techniques has been the pivotal to the emergence of EEG field over the last couple of years. More importantly, researchers on BCI technology has succeeded in theirs initial goals of providing a proof-of-concept of their potential benefits for people affected by neuromuscular diseases [10]. Nowadays, research is moving toward gaining and capturing the attention of general public to whom this technology are still of the domain of science fiction

1.2. Literature Review

This section will provide a comprehensive overview of BCI systems, theirs evolution and the different groups currently leading the research within the field along with theirs achieved works. Then present the origin of EEG signal coupled with how they are recorded and their characteristics, finally some the technique used in the design of BCI will be presented.

The history of BCI ignites back in 1924 during the First World War with Hans Berger who discovered the bioelectrical signal generated from brain electroencephalographic (EEG) activity. He is credited to be the first person to record human brain activity by means of EEG signal. Berger was able to identify oscillatory activity by analysing EEG traces, which was then named after him as Berger's wave or the alpha wave which occurs around (8–13 Hz). Unfortunately, Berger's first recording device was very rudimentary. It was made of silver foils attached to the patient's head using rubber bandages, these foils were then connected to a Lippmann capillary electrometer with disappointing results [12]. It is only in 1970 that the term BCI was coined by Prof Vadal to denote the phenomenon of brain bioelectrical signal for human computer interface. Since then, intensive studies were carried out in order to decode and understand these signals originating from brain activity. However back in 70’s, this task was thought to be almost impossible because several factors [13] among which: the EEG is a very complex signal since its contents reflects not only the activity of certain region of the brain but also the electrical activity of trillions of synapses in the cortex [11]. In addition, EEG-based BCI require a real-time capturing, and processing of the signal, the technology required for that was still at a very primitive stage and was extremely expensive. Finally, the capabilities and hence potentials usages of an EEG-based BCIs were still highly underestimated [12].

In the recent years, the ongoing development in applied neurophysiology science coupled with the advancement of the EEG signal processing method has contributed to the fast growing emergent of BCI to a level which was unimaginable 60 years ago. A huge variety of porotype BCI systems have been implemented for both medical and research purposes with satisfactory performances results.

The technology has proven to be of high benefit to people suffering neuromuscular disease in the sense that it enable them to gain back some sense of autonomy via controlling objects or displaying letters on a virtual keyboard hence not only does BCI devices improve their quality of life but they also reduce some of the assistive care[1].

1.2.1. EEGs

Electroencephalogram is the workhorse medical procedure used to record the average electrical activity at the different locations of human brain. In order words, it measures the ionic current flows resulting synaptic excitation of large group of neurones in the cerebral cortex [6]. In the medical context, EEG refers to the recording of the tiny brain’s spontaneous electrical activity over a short period of time know as impulses. The brain activity is recorded using dry electrodes imbedded with saline solution which are placed on the user’s scalp according to the 10-20 system [6] illustrated in figure 1:

Figure 1: EEG Electrodes Placement following the 10-20 system [6]

The name associated with each electrodes position refers to the anatomical region of the head cortex above which they are situated; Fp refers to the frontoolar region, F refers to frontal, T refers to temporal, C refers to central, P refers to parietal, O refers to occipital and A refers to auricular while G stands for the ground electrodes. The right and left side of the brain are respectively denoted by the associated even and odd number [3]. The EEGS signal recorded by the electrodes often range between 10μV to 100μV in amplitude and they often suffer from lots of noise interference/ artefacts [3], which is the reason why sophisticated recording devices need to be employed along with an adequate processing technique in order to achieve a satisfactory BCI design system.

1.2.2. Machine Learning

Machine learning can simply be defined as the process of adapting a certain system to a specific user base on some pre-recorded data. The idea of learning from data has been used in multiple applications where the system was non-deterministic, but there are data’s available to analytically describe how the system operates. In the context of BCI system, the user’s brain activity (encoded data’s) have to be translate into a certain command for a computer or electronic devices. An intuitive approach will be to map the every recorded EEG signal to the mental action that caused its genesis. The goal being to look at the distribution of some discriminative information within the signal and then estimate the adequate translation rule. For instance, a closely related methodology is used in [14]  during the early stage of BCI technology by JR Wolpaw, where the user was able to move a cursor up and down based on variation in amplitudes of its brain waves. The correlation between the each cursor motion and the corresponding brain signal amplitude was made manually by a mathematical function. However, JR Wolpaw acknowledge the fact that it even if a unique feature of the recorded EEG signal is used during the mapping process, it very arduous for a human to specify an optimal mapping between commands and signals. In addition, brain signal are user’s specific .i.e. they vary quite a lot from one individual to another, this means the translation rule will have to be update for every new user which is a very tedious and demanding exercise.

A solution uses in modern BCI design to automate the translation rule is via computer aids machine learning algorithm [15]. First, the user’s EEG signal is recorded when tasked to perform some prescribed actions, the data’s are then grouped in two classes using a computer algorithm during a process commonly referred as training phase in BCI design. In this project two types of supervised machine learning algorithm were in implemented as part of the proof of concept BCI design: the support vector machine (SVM) and the linear discriminant analysis (LDA).

1.2.3. BCI

A BCI system can be considered as an artificial window to the brain, use to measure the user’s brain oscillatory activity associated with his/her intents and translated them into manful signals [1]. Since the initial reference in the 70s by prof. Vadal, BCI has been a fast growing technology and research topic with a primarily focused on neuroprosthesis applications; some of the pioneering works and publications credited to Fetz team carried on animals, showed for the first time that monkeys could control a robot arm with theirs brain. Subsequently, a first EEGs based system was tested on patient suffering from amyotrophic lateral sclerosis (ALS) by Birbaumer et al. in 1999 highlighted the fact that those patients could use BCI system to control spelling devices to interact with theirs surroundings [16]. In 2005, professor Jonathan Wolpaw was acknowledge to develop the first modern BCI with electrodes positioned on the surface of the scalp.

Following up to that, the popularity of BCI technology kept rising surprisingly with related publications, conference talks, products, etc. being made every years. Currently, there are up to 200 different research groups actively working within the field [17]; this has not only led to a much better understanding of human brain activity and rapid advancement in brain signals processing methods, but also contributed to expand the usability ranges of this devices.

Based on how the brain signal is being recorded, brain computer interface (BCI) system are regrouped in the following categories [18]:

 Invasive BCIs where the electrodes are inserted directly into the user’s brain by critical surgery. The technique use is known as electro-corticogram; despite having the good quality brain signal resolution there a significant risks associated

 Non-Invasive system, here the electrodes are placed on the surface of the user’s scalp; they are safer but only provide a weak brain signals. This include EEG (electroencephalogram), NIRS (near infrared-spectroscopy), and fMRI (functional magnetic resonance imaging) etc.…

 P300 based system ( record event related potentials that occurs between 250-500ms ), steady-state visual evoked potential (SSVEP) systems and motor imaginary based on the BCI principle used

 There are several others characterisation criteria used to differentiate BCI system such as speed and accuracy, the training time, synchronous and asynchronous system etc.…

Some of the research groups leading the research on BCI technologies are presented in table 1 below along with a summary of their related works and publications.

Research Group BCI Project Publication & Year

Christine E. King, An H. Do, Zoran Nenadic, from University of California, Irvine, USA The feasibility of a BCI functional electrical stimulation system for the restoration of overground walking after paraplegia :

A man who has been paralysed following a cord injury has been able to walk a distance of 3.66m by the only mean of his brain power. The patient regains some mobility using an EEG-based system, which captured electrical signal from his brain, translated these into control signal which then passed to electrodes placed around his knees to creates the walking movement

Journal of NeuoEngineering and Rehabilitation, 2015; 12 (1)

H. Zhang, R. Chavarriaga, Z. Khaliliardali from the Chair in Brain-Machine Interface research group at EPLF EEG-Based Decoding of Error Brain Activity in a Real-world Driving Task

This study expands the possible application of BCIs in car driving, by integrating them into driving assistant system to predict the driver level of attention, estimate his/her mental workload or anticipate driver intention

Journal of Neural Engineering, Vol.12, num. 6, 2015.

Jing Guo, Shangkai Gao, Bo Hong. From university of Japan

An active auditory BCI for intention expression in Locked in

The presented BCI system uses random sequence of sound options to enable the user’s (severely paralysed patients) to express his/her intentions

Annual BCI Research Award, 2010.

Setare Amiri, Ahmed Rabbi, Leila Azinfar and Reza Fazel-Rezai. From University of North Dakota, USA A Review of P300, SSVEP and Hybrid P300/SSVEP BCIs Systems

The paper lists some recent signals processing method, features extraction methods and classification techniques along with how they can be combined to produce a new trend of hybrid BCI system with higher accuracy, reliability and data transferred rate Brain –Computer Interfaces Systems- Recent Progress and Future Prospects.

Lyn M.McCane, Susan M. Heckman, Dennis J. McFarland from the Laboratory of Neural Injury and Repair, Wadsworth Centre, New York State, USA P300-Based BCI event related potentials: people with amyotrophic lateral sclerosis vs age matched controls

The present study compare the performances of P300 BCI speller on healthy person and paralysed patient via an offline analyses in which the subject are asked to target a letter on flashing matrix. The paper claims that there were almost no different between the results achieved by disables patient and healthy volunteer

Clinical Neurophysiology, 20015.

Table 1: Current Research Trend in Brain Computer Interface

A good amount of  work has been in last decades on BCI technologies, while the first task of acquiring the input data (brain signals) seems to have been well advanced, the understanding, processing and extraction of important information has proven to be more difficult leading to the implementation of a multitude of methods and features extraction techniques

1.2.4. Methods and Techniques

There are several approach used to measure brain activity or its response to an external stimuli; the choice of the methods relies on the intended usage or application of the BCI system.

Quantitative Electroencephalography (QEEG) processing method uses mathematical techniques like Fast Fourier Transform and Wavelet Transform analysis to analysis the recorded brain signals from multi- channels electroencephalogram electrodes placed at the surface of the user’s scalp [18]. This analysis observes the dynamic changes happening in the user's cognitive state when asked to think about certain actions.

Electrocorticography (ECoG), the electrodes that are used to monitor the brain activity are directly placed under the user’s skull via risky procedure. However, the ECoG signals have a higher signal to noise ratio, less sensitive to artefacts along with a width amplitude and good resolution. BCI system based on ECoGs provide a better data transmission rate [18].

Magnetoencephalography (MEG) technique records the change in magnetic field generated by charged ions excited within the neurons of the user’s (patient) brain activity. The use of MEG in BCI systems has proven to have some great advantages such as a higher temporal resolution of the recorded brain signal, making them more reliable for BCI communication [18]. However, due theirs relative big sizes, they cannot be used to build mobile BCI devices; reason why they often find application mainly in the medical and research filed.

Functional Magnetic Resonance Imaging (fMRI) is a novel technique based on magnetic resonance imaging, to establish a map of the patient brain activity under stimuli. It measures the changes in blood oxygenation and flow that occurs in response to a mental tasks [1]. When the different regions of the brain associated with as specific cognitive process are successfully localise, the corresponding brain signals is the recorded. fMRI is of particular interest in measuring physiological diseases

...(download the rest of the essay above)

About this essay:

This essay was submitted to us by a student in order to help you with your studies.

If you use part of this page in your own work, you need to provide a citation, as follows:

Essay Sauce, . Available from:< > [Accessed 25.05.20].