Chapter 2 2


Chapter 2
2.1 Epilepsy and neurological condition
Epilepsy and neurological condition Epilepsy basically is a neurological disorder which starts with sudden neurological sensory disturbance and can create loss of consciousness or convulsions associated with abnormal electrical activities of brain. Epilepsy is the fourth most neurological disorder. It affects the people from all ages and can causes unpredictable seizures likes burn, submersion injuries, falls and many more. Epilepsy can occur at any time without giving any signal or symptom. So patients cannot prepare or protect this problem. It is reviewed that children with epilepsy are the most vulnerable for submersible injuries. Though epilepsy is a central nervous system disorder during epilepsy, brain creates various abnormal signals causing unusual sensations and other abnormal activities ADDIN EN.CITE <EndNote><Cite><Author>Jacoby</Author><Year>2005</Year><RecNum>2562</RecNum><DisplayText>1</DisplayText><record><rec-number>1</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459558″>1</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Jacoby, Ann</author><author>Snape, Dee</author><author>Baker, Gus A.</author></authors></contributors><titles><title>Epilepsy and social identity: the stigma of a chronic neurological disorder</title><secondary-title>The Lancet Neurology</secondary-title></titles><periodical><full-title>The Lancet Neurology</full-title></periodical><pages>171-178</pages><volume>4</volume><number>3</number><section>171</section><dates><year>2005</year></dates><isbn>14744422</isbn><urls></urls><electronic-resource-num>10.1016/s1474-4422(05)70020-x</electronic-resource-num></record></Cite></EndNote>1.

During the last few decades many resources are placed based on EEG for determining the neurological condition during the epilepsy. It is difficult to find the brain condition when epilepsy occurs even for the experienced physicians. During the epilepsy, seizures can takes different format and can affect the people in different way. The normal activities can take place during epilepsy when the brain is activated by seizures discharge which is known as “Electrical storms” in brain. There are three stages of epileptic seizure beginning, middle and ending. But all doesn’t feel all the stages in the same way ADDIN EN.CITE ;EndNote;;Cite;;Author;Valerie Jewells;/Author;;Year;2014;/Year;;RecNum;2581;/RecNum;;DisplayText;2;/DisplayText;;record;;rec-number;2;/rec-number;;foreign-keys;;key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459558″;2;/key;;/foreign-keys;;ref-type name=”Journal Article”;17;/ref-type;;contributors;;authors;;author;Valerie Jewells, Hae Won Shin;/author;;/authors;;/contributors;;titles;;title;Review of Epilepsy – Etiology, Diagnostic Evaluation and Treatment;/title;;secondary-title;International Journal of Neurorehabilitation;/secondary-title;;/titles;;periodical;;full-title;International Journal of Neurorehabilitation;/full-title;;/periodical;;volume;01;/volume;;number;03;/number;;dates;;year;2014;/year;;/dates;;isbn;23760281;/isbn;;urls;;/urls;;electronic-resource-num;10.4172/2376-0281.1000130;/electronic-resource-num;;/record;;/Cite;;/EndNote;2.

2.1.1 Beginning
During the beginning some can aware of the beginning of the seizure on the other hand some may not. Some people may experience feelings, sensations or changes in behavior before a day or hours of a seizure which is not a part of seizure but can warn people that something may come. This symptom is known as “Prodeome”. Some may also feel an aura or warning which is the first symptom of seizure and this is considered as a part of the seizure. Few times an aura can recognize by change of feelings and behavior but most of the time it is indescribable feelings ADDIN EN.CITE ;EndNote;;Cite;;Author;Angalakuditi;/Author;;Year;2011;/Year;;RecNum;2582;/RecNum;;DisplayText;3;/DisplayText;;record;;rec-number;3;/rec-number;;foreign-keys;;key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459558″;3;/key;;/foreign-keys;;ref-type name=”Journal Article”;17;/ref-type;;contributors;;authors;;author;Angalakuditi, M.;/author;;author;Angalakuditi, N.;/author;;/authors;;/contributors;;auth-address;Georgia State University, Atlanta, GA;;/auth-address;;titles;;title;A comprehensive review of the literature on epilepsy in selected countries in emerging markets;/title;;secondary-title;Neuropsychiatr Dis Treat;/secondary-title;;/titles;;periodical;;full-title;Neuropsychiatr Dis Treat;/full-title;;/periodical;;pages;585-97;/pages;;volume;7;/volume;;edition;2011/10/18;/edition;;keywords;;keyword;economics;/keyword;;keyword;emerging markets;/keyword;;keyword;epidemiology;/keyword;;keyword;epilepsy;/keyword;;keyword;guidelines;/keyword;;keyword;seizures;/keyword;;keyword;treatment patterns;/keyword;;/keywords;;dates;;year;2011;/year;;/dates;;isbn;1178-2021 (Electronic) 1176-6328 (Linking);/isbn;;accession-num;22003298;/accession-num;;urls;;related-urls;;url;https://www.ncbi.nlm.nih.gov/pubmed/22003298;/url;;/related-urls;;/urls;;custom2;PMC3191871;/custom2;;electronic-resource-num;10.2147/NDT.S24966;/electronic-resource-num;;/record;;/Cite;;/EndNote;3. The aura can also occur alone and may be called a focal onset aware seizure, simple partial seizure or partial seizure without change in awareness.

2.1.2 Middle
The middle stage of epileptic seizure may call ictal phase. The time period from the first symptom to the end of the seizure is called middle period. This is related with the electrical activities of the brain. During this period some common symptoms are observed like loss of awareness, confused, distracted, day dreaming, unable to hear, blurry vision, numbness, tingling or electrical shock like feeling in the body etc. During this period some physical changes occur like difficulty of talking, repeated of blinking eyes, lack of movement of muscle, tremors, twitching, change of skin color, pupils may dilated or appear larger than regular, difficulty in breathing and mostly heart racing etc.

2.1.3 Ending
When seizure ends the postictal phase is start. It is basically the recovering stage of brain and body. Sometimes a fast recovery is possible for some kind of people and sometimes it may take hours to recover the usual feelings. There are always some symptoms for each stage of the epileptic seizure. Like the other stages ending has some common symptoms too. Like slow response or not able to response at all right away, sleepy, confused, scared, anxious, frustrated, ashamed etc. Besides these some physical changes also shows like injuries, tired, exhausted, headache and other pain, nausea, thirsty, losing control of bowel or bladder etc.

2.2 Brain Signals
Brain signals are the electrical response of the neuron to neuron communication during thoughts, emotions and behaviors. Brain signals are produced by synchronized electrical pulsed from numerous number of neuron communicating each other. Our brain signals change according to the feelings and our activities. When slow brain signals are dominant we feel tired, slow, sluggish or dreamy. The higher frequency signals are dominant when we feel wired and hyper besides when we think about love or lust higher frequency signal may generate. According to the strength of the signal they are classified ADDIN EN.CITE ;EndNote;;Cite;;Author;Kaur;/Author;;Year;2015;/Year;;RecNum;2584;/RecNum;;DisplayText;4;/DisplayText;;record;;rec-number;4;/rec-number;;foreign-keys;;key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459559″;4;/key;;/foreign-keys;;ref-type name=”Conference Proceedings”;10;/ref-type;;contributors;;authors;;author;J. Kaur;/author;;author;A. Kaur;/author;;/authors;;/contributors;;titles;;title;A review on analysis of EEG signals;/title;;secondary-title;2015 International Conference on Advances in Computer Engineering and Applications;/secondary-title;;alt-title;2015 International Conference on Advances in Computer Engineering and Applications;/alt-title;;/titles;;pages;957-960;/pages;;keywords;;keyword;bioelectric potentials;/keyword;;keyword;electroencephalography;/keyword;;keyword;medical signal processing;/keyword;;keyword;statistical analysis;/keyword;;keyword;EEG signal analysis;/keyword;;keyword;brain;/keyword;;keyword;electrical bustle;/keyword;;keyword;electrical voltage;/keyword;;keyword;signal processing;/keyword;;keyword;statistical approaches;/keyword;;keyword;time-varying nonstationary signals;/keyword;;keyword;Brain modeling;/keyword;;keyword;Electrodes;/keyword;;keyword;Mathematical model;/keyword;;keyword;Principal component analysis;/keyword;;keyword;Wavelet transforms;/keyword;;keyword;EEG signals;/keyword;;keyword;analysis;/keyword;;keyword;methods;/keyword;;/keywords;;dates;;year;2015;/year;;pub-dates;;date;19-20 March 2015;/date;;/pub-dates;;/dates;;urls;;/urls;;electronic-resource-num;10.1109/ICACEA.2015.7164844;/electronic-resource-num;;/record;;/Cite;;/EndNote;4.

2.2.1 Infra-low (;.5Hz)
Infra-low signals are our higher brain functional signals thought as the basic cortical rhythm. Very little information is available about this type of signal. Their slower nature makes them difficult to identify and measure PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5NaXRyYTwvQXV0aG9yPjxZZWFyPjIwMTg8L1llYXI+PFJl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ADDIN EN.CITE PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5NaXRyYTwvQXV0aG9yPjxZZWFyPjIwMTg8L1llYXI+PFJl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ADDIN EN.CITE.DATA 5.

2.2.2 Delta Signal (.5-3 Hz)
Delta signals are also slow but loud like dram beat. They travel slow but penetrating. This type of signal is generated during the deepest meditation and dreamless sleep. Healing, regeneration and empathy is the creator of such signals in the brain.

Fig1: Delta Wave signal
2.2.3 Theta Signal (3-8 Hz)
Theta wave signals are very important signal. This is generated during our learning, memory and intuition. When we are in dream, vivid imagery, intuition and information beyond our normal conscious awareness theta wave signal generate. It is where we store our fears, troubled history

Fig2: Theta Signal
and nightmares. It is basically seen in young children and older children and adults may see this in drowsiness or arousal. Excess theta for age represents abnormal activities of the brain. It can be seen in generalized distribution in diffuse disorder or metabolic encephalopathy or deep midline disorders or some instances of hydrocephalus.

2.2.4 Alpha Signals (8-13 Hz)
Hans Berger realized this signal very firstly and named this as alpha wave or signal. Alpha is the signal having frequency range of 8-13 Hz. It is higher in amplitude on the dominant side, seen in the posterior regions of the head on both side that’s why it is called posterior rhythm or posterior alpha rhythm. Alpha is related to resource allocation in the cortex, and is produced as a result of

Fig3: Alpha signal
a resonance process between the thalamus and the cortex. If we consider the thalamus the gateway to the cortex, alpha can be thought of as the mechanism by which the sensory gate to the cortex can be closed.

2.2.5 Beta Signals (13-38 Hz)
Beta signals are operating within 13-38 Hz frequencies. It is a fast wave signal. Basically Beta signals represent our normal waking state of consciousness when we are in attention and it is directed at cognitive tasks and outside world. This type of signals are dominated when we are alert, attentive and engages in problem-solving, decision making and focused mental activities.

Fiq4: Beta signal
Beta signals can divide into 03 different stages. Low-beta signal (12-15 Hz) which is thought to be “fast idle”, or musing thought, Beta signal (15-22 Hz) is high-engagement and actively figuring things out. And finally High Beta (22-38 Hz) is a high frequency signal which generate in complex thought, integrating new experiences, high anxiety or excitement ADDIN EN.CITE <EndNote><Cite><Author>Zainuddin</Author><Year>2014</Year><RecNum>2592</RecNum><DisplayText>6</DisplayText><record><rec-number>6</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459559″>6</key></foreign-keys><ref-type name=”Conference Proceedings”>10</ref-type><contributors><authors><author>B. S. Zainuddin</author><author>Z. Hussain</author><author>I. S. Isa</author></authors></contributors><titles><title>Alpha and beta EEG brainwave signal classification technique: A conceptual study</title><secondary-title>2014 IEEE 10th International Colloquium on Signal Processing and its Applications</secondary-title><alt-title>2014 IEEE 10th International Colloquium on Signal Processing and its Applications</alt-title></titles><pages>233-237</pages><keywords><keyword>data acquisition</keyword><keyword>electroencephalography</keyword><keyword>feature extraction</keyword><keyword>signal classification</keyword><keyword>EEG brainwave signal classification technique</keyword><keyword>alpha signals</keyword><keyword>beta signals</keyword><keyword>data sampling</keyword><keyword>data session</keyword><keyword>electroencephalograph signal</keyword><keyword>functional electrical stimulation</keyword><keyword>Classification algorithms</keyword><keyword>Conferences</keyword><keyword>Electrodes</keyword><keyword>IEEE Engineering in Medicine and Biology Society</keyword><keyword>Indexes</keyword><keyword>Electroencephalograph (EEG)</keyword><keyword>Intelligent Classification</keyword><keyword>alpha</keyword><keyword>beta</keyword><keyword>signal processing</keyword></keywords><dates><year>2014</year><pub-dates><date>7-9 March 2014</date></pub-dates></dates><urls></urls><electronic-resource-num>10.1109/CSPA.2014.6805755</electronic-resource-num></record></Cite></EndNote>6.

2.2.6 Gamma Signals (38-100 Hz)
Gamma brain signals have the highest frequencies of any brainwave, oscillating between 38-100 Hz. Peak concentration and high levels of cognitive functioning are associated with gamma signals. There are different levels of gamma signals and has different functionality. Learning

Fig5: Gamma Signal
difficulty, impaired mental processing and limited memory have been linked with low levels of gamma signals. Besides this high gamma activity is correlated with high IQ, compassion, excellent memory and happiness. Gamma signal and theta signals works together to recruit neurons which stimulate local cell column activity.

2.3 EEG & It’s characteristics
An electroencephalogram (EEG) is basically a test which is used to evaluate brain electric activities. Brain has numerous number of brain cells which is called neurons. These brain cells communicate with each other through electric impulse. EEG is used to detect this potential problems associated with this activity.

Electroencephalography (EEG) is an electrophysiological monitoring method to record electrical activity of the brain. It is typically noninvasive, with the electrodes placed along the scalp, although invasive electrodes are sometimes used such as in electro-corticography. EEG measures voltage fluctuations resulting from ionic current within the neurons of the brain. In clinical contexts, EEG refers to the recording of the brain’s spontaneous electrical activity over a period of time, as recorded from multiple electrodes placed on the scalp.

Fiq: Various portion of brain
Diagnostic applications generally focus either on event-related potentials or on the spectral content of EEG. The former investigates potential fluctuations time locked to an event like stimulus onset or button press. The latter analyses the type of neural oscillations (popularly called “brain waves”) that can be observed in EEG signals in the frequency domain.

EEG is most often used to diagnose epilepsy, which causes abnormalities in EEG readings. It is also used to diagnose sleep disorders, depth of anesthesia, coma, encephalopathy, and brain death. EEG used to be a first-line method of diagnosis for tumors, stroke and other focal brain disorders, but this use has decreased with the advent of high-resolution anatomical imaging techniques such as magnetic resonance imaging (MRI) and computed tomography (CT). Despite limited spatial resolution, EEG continues to be a valuable tool for research and diagnosis. It is one of the few mobile techniques available and offers millisecond-range temporal resolution which is not possible with CT, PET or MRI.

An EEG tracks and record brain wave patterns. Small metal disc called electrode is used to track the brain wave attached with scalp with wires. Where the electrode analyzes the electric activities of brain and sent signals to a computer by wires that record the results.

In the EEG recording the electrical impulse looks like wavy lines with various peaks. From this peaks any abnormal brain activity can detect. According to the amplitude, frequency of the wave the name of the signal is different and their characteristics are different. From a little variation of this signal or wave the disorder or abnormality of the brain can detect. EEG is basically use for detecting problems like seizure disorder such as epilepsy, head injury, brain tumor, encephalopathy which is a disease that causes brain dysfunction, memory problem, sleep disorders, stroke, dementia etc.

2.4 Seizure detection using EEG
For the purpose of detecting seizure EEG is a great medium. EEG is easy safe and almost accurate. There are different ways to detect seizure by EEG. Some ways are given below
2.4.1 Feature Extraction by Spectral Decomposition
In seizure detection spectral decomposition is a great way for time domain signal analysis. In EEG feature extraction process is determine by some algorithms. Among these one of the most widely used algorithms is furrier series analysis. The classical furrier transformation is the expression of signal as a linear combination of complex exponentials. But for no-linear situation is required of modification. Short time furrier transformation (STFT) is the way to use in non-linear signal analysis. The short time furrier transformation handles the non-linear and non-stationary signals by windowing process. This process is successfully employed in seizure detection by EEG PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5FbGlzaGE8L0F1dGhvcj48WWVhcj4yMDE3PC9ZZWFyPjxS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ADDIN EN.CITE PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5FbGlzaGE8L0F1dGhvcj48WWVhcj4yMDE3PC9ZZWFyPjxS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ADDIN EN.CITE.DATA 7. The inability of complex exponentials to model real world problems lead to the formulation of generalized Fourier transforms and fractional Fourier transforms. The generalized Fourier transform uses functions other than complex exponential. Bessel Functions; while the fractional Fourier transforms are based on concepts of fractional calculus that extends fractional dimensions to transform methods. The applications of fractional Fourier transform to a broad class of bio-medical signals have been proposed. The non-stationary behavior of signals leads to the application of time-frequency analysis (TF analysis).This class of methods attempts to capture the non-stationary behavior of time series, and thus has been applied to the seizure detection problem. One of the most important TF methods, the wavelet transform, tries to optimize the choice of window size by introducing small window sizes for high frequencies and large windows for low frequencies. It also incorporates the idea of diversifying the basic functions by the introduction of several mother wavelets. Classical formulations of wavelet transform have been widely used for seizure detection. A variation of the time-frequency distributions, known as the S-transform, has also been used to solve this problem. Although it yields better signal clarity than the wavelet transform, but is computationally costly. All the time-frequency methods are subject to the Heisenberg uncertainty principle that governs the trade-off between the accuracy in measurements of time and frequency. Empirical mode decomposition (EMD) is a heuristic method of signal analysis that has proven to be useful for non-linear and non-stationary signals.
2.4.2 Time domain Feature Extraction
Besides the spectral method a lot of temporal method is used to detect seizure by EEG. One of them is time domain feature extraction. In this method of detection the aim is to try and quantify the different states of brain underlying the probabilistic distribution. In order to quantify the coherence of the underlying states, a number of methodologies have been used to build estimators of phase coherence. The phase of a real-valued time series can be constructed in a number of ways. The most common of them is Taken’s embedding, which is similar to windowing (using sliding rectangular windows), and has been at the heart of many seizure detection techniques. Another popular method for the construction of phase space, is to transform the signal using complex basis vectors, e.g., complex wavelets or complex exponentials, etc. This transformation is followed by phase construction from this resulting analytic signal. This method has also been the focus of numerous recent methods employed for seizure detection. Another important method of quantifying the coherence of the states is to calculate the entropy of the time series. Entropy is the measure of uncertainty of a data stream and if the underlying states are correlated, the resulting time series will be more ordered thus having lower entropy. There are numerous variations of the concept of entropy, including Shannon entropy, approximation entropy, permutation entropy and fuzzy entropy. All of these have been employed for seizure detection. ADDIN EN.CITE ;EndNote;;Cite;;Author;Ahmad;/Author;;Year;2016;/Year;;RecNum;2520;/RecNum;;DisplayText;8;/DisplayText;;record;;rec-number;8;/rec-number;;foreign-keys;;key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459559″;8;/key;;/foreign-keys;;ref-type name=”Book”;6;/ref-type;;contributors;;authors;;author;Ahmad, Muhammad;/author;;author;Saeed, Maryam;/author;;author;Saleem, Sajid;/author;;author;Kamboh, Awais;/author;;/authors;;/contributors;;titles;;title;Seizure Detection using EEG:A survey of different Techniques;/title;;/titles;;dates;;year;2016;/year;;/dates;;urls;;/urls;;electronic-resource-num;10.1109/ICET.2016.7813209;/electronic-resource-num;;/record;;/Cite;;/EndNote;8
2.5 Methods
In 1971 the Indian Epilepsy Association (IEA) was registered for the first time to work for raising epilepsy awareness and provided rehabilitation as well. During the mid of the last century in 1960’s to 1970’s many researchers were trying to develop the first automated seizure detection and prediction system using computerized method.
The approaches for predicting and detection of seizure were based on some major sectors. Like
Neural network based
Wavelet-based
Fourier transformation based
Classifier based
Other techniques
Besides these many researchers have used and worked with advanced computerized algorithms and new technologies implemented for seizure detection and prediction. Here we will review only the major contribution in the history of seizure prediction and detection
2.5.1 Neural network based approaches
Neural network based seizure prediction and detection technique is the complicated and latest technique which is now used in a large number of new researches. In 1996 Weng and Khorasani proposed the latest technique based on neural network. Their introduction algorithm was based on the class of neuron-generating strategies. A simple multilayered perceptron type network was used and then stabilized error function was used to make decision whether there is any need to generate new neuron or not. This algorithms was observed to have improved results about 60%-70% improved results over the other techniques adopted during that time ADDIN EN.CITE ;EndNote;;Cite;;Author;Weng;/Author;;Year;1996;/Year;;RecNum;2580;/RecNum;;DisplayText;9;/DisplayText;;record;;rec-number;9;/rec-number;;foreign-keys;;key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459559″;9;/key;;/foreign-keys;;ref-type name=”Journal Article”;17;/ref-type;;contributors;;authors;;author;Weng, W.;/author;;author;Khorasani, K.;/author;;/authors;;/contributors;;titles;;title;An Adaptive Structure Neural Networks with Application to EEG Automatic Seizure Detection;/title;;secondary-title;Neural Networks;/secondary-title;;/titles;;periodical;;full-title;Neural Networks;/full-title;;/periodical;;pages;1223-1240;/pages;;volume;9;/volume;;number;7;/number;;keywords;;keyword;Adaptive structure networks;/keyword;;/keywords;;dates;;year;1996;/year;;pub-dates;;date;1996/10/01/;/date;;/pub-dates;;/dates;;isbn;0893-6080;/isbn;;urls;;related-urls;;url;http://www.sciencedirect.com/science/article/pii/0893608096000329;/url;;/related-urls;;/urls;;electronic-resource-num;https://doi.org/10.1016/0893-6080(96)00032-9;/electronic-resource-num;;/record;;/Cite;;/EndNote;9.
In 1975 Sam et al. developed a warning system for predicting the epileptic seizure that was worked based on the pattern recognition principles. Here the first miniature size device was invented for patients to use in advanced warning system related to any imminent seizure.
Then in 1999, Christopher et al. presented a fully automated detection system for seizure detection using EEG activities of the brain and tested over 43 patients on that time. That proposed technique was based on a popular artificial neural network (ANN) that is well known as self-organizing feature map (SOFM). In this technique the overall sensitivity was 82% with the huge reduction in false detection rate.

Then Hen et al. (2002) developed a robust system which helps to combine the multiple signal-processing techniques in the ANN, wavelet transform, integrated adaptive filtering, multistage scheme, and expert system. It is developed with two major stages: the first one is basic preliminary-type screening stage where data reduction occurs very significantly and the second one is the analytical stage. The sharp spikes were used to design the seizure detection system earlier but this latest technique takes the slower signals of the EEG as consideration. As a result the authors provided a result that gave 90% and 90.47% of accuracy and decreasing false detection rate respectively.
After this many new proposed techniques were developed using the ANN. Maryann et al. proposed another technique using the pattern recognition system with ANN that could classified the subject on the basis of whether they are affected by the epilepsy or not. In this method by the use of confusion matrix and the receiver-operating characteristics (ROC) it becomes easier to find a healthy and unhealthy subject. This method was able to get the form of sensitivity up to 62.5% and the specificity with practical value as 90.47%. This was obtained by Abdulhamit et al. in 2005 ADDIN EN.CITE ;EndNote;;Cite;;Author;Subasi;/Author;;Year;2005;/Year;;RecNum;2615;/RecNum;;DisplayText;10;/DisplayText;;record;;rec-number;10;/rec-number;;foreign-keys;;key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459560″;10;/key;;/foreign-keys;;ref-type name=”Journal Article”;17;/ref-type;;contributors;;authors;;author;Abdulhamit Subasi;/author;;/authors;;/contributors;;titles;;title;Epileptic seizure detection using dynamic wavelet network;/title;;secondary-title;Expert Syst. Appl.;/secondary-title;;/titles;;periodical;;full-title;Expert Syst. Appl.;/full-title;;/periodical;;pages;343-355;/pages;;volume;29;/volume;;number;2;/number;;dates;;year;2005;/year;;/dates;;isbn;0957-4174;/isbn;;urls;;/urls;;custom1;1707585;/custom1;;electronic-resource-num;10.1016/j.eswa.2005.04.007;/electronic-resource-num;;/record;;/Cite;;/EndNote;10. Vladimir et al. (2007) proposed the latest approach for detection of epileptic seizures using chaos theory, correlation dimension, or namely determination largest Lyapunov’s exponent for scalp EEG signals.

Forrest et al. (2008) focused on the development of highly automated diagnosis system for epileptic patients that can work on the basis of inter-ictal EEG data sets. During this research activity, three classes of EEG signal features were formed and then a probabilistic neural network (PNN) fed was developed using this feature-based signal information. This was able to obtain around 99.3% of overall accuracy. On the other hand, for seizure focus localization the accuracy was recorded to be 76.5% and for patient monitoring, the value of accuracy was obtained to be 96.7%.

Beside these Alexandros et al. (2009) proposed a time–frequency analysis for classification of epileptic seizures with final accuracy of 99.28% ADDIN EN.CITE <EndNote><Cite><Author>Tzallas</Author><Year>2009</Year><RecNum>2589</RecNum><DisplayText>11</DisplayText><record><rec-number>11</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459560″>11</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Tzallas, A. T.</author><author>Tsipouras, M. G.</author><author>Fotiadis, D. I.</author></authors></contributors><auth-address>Unit of Medical Technology and Intelligent Information Systems, Department of Material Science and Technology, University of Ioannina, Ioannina 45110, Greece. [email protected]</auth-address><titles><title>Epileptic seizure detection in EEGs using time-frequency analysis</title><secondary-title>IEEE Trans Inf Technol Biomed</secondary-title></titles><periodical><full-title>IEEE Trans Inf Technol Biomed</full-title></periodical><pages>703-10</pages><volume>13</volume><number>5</number><edition>2009/03/24</edition><keywords><keyword>Bayes Theorem</keyword><keyword>Electroencephalography/*methods</keyword><keyword>Epilepsy/classification/*diagnosis</keyword><keyword>Fourier Analysis</keyword><keyword>Humans</keyword><keyword>Logistic Models</keyword><keyword>Neural Networks (Computer)</keyword></keywords><dates><year>2009</year><pub-dates><date>Sep</date></pub-dates></dates><isbn>1558-0032 (Electronic) 1089-7771 (Linking)</isbn><accession-num>19304486</accession-num><urls><related-urls><url>https://www.ncbi.nlm.nih.gov/pubmed/19304486</url></related-urls></urls><electronic-resource-num>10.1109/TITB.2009.2017939</electronic-resource-num></record></Cite></EndNote>11. In 2010 Gayatri et al. proposed another method based on ANN and that was named as Ekman neural network. In this method it was observed the value of approximated entropy used to decay sharply when epileptic seizures occurred. Pravin Kumar et al. (2009) used features based on entropy to discriminate normal and abnormal EEG data sets. They gave major preference to three nonlinear features such as spectral entropy, sample entropy, and wavelet entropy so that quantitative features from input EEG signal can be recorded. After this feature extraction process, two unique neural network models were applied: radial basis network and recurrent Elman Network in order to classify healthy and unhealthy signals. In 2013 Pradipta et al. proposed another method based on EEG signals using neural network and discrete wavelet transform (DWT) to detect epileptic disease. Here the ultimate value was recorded as average specificity 99.19%, selectivity=91.4% and sensitivity=91.29%.

Gamze et al. (2015) developed a MATLAB Graphical User Interface (GUI) based automated diagnostic system for the epilepsy disease. For classification of EEG, three different methods of ANNs, Cascade, Elman, and Feed Forward Back-propagation, are used. From the obtained experimental results, it is observed that the proposed diagnostic system can accomplish accuracy of 98.3% and the GUI-based interface is much easier to use for medical personnel ADDIN EN.CITE <EndNote><Cite><Author>Bozkurt</Author><Year>2015</Year><RecNum>2588</RecNum><DisplayText>12</DisplayText><record><rec-number>12</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459560″>12</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Gamze Dogali Çetin Özdemir Çetin Mehmet Recep Bozkurt</author></authors></contributors><titles><title>The Detection of Normal and Epileptic EEG Signals using ANN Methods with Matlab-based GUI</title><secondary-title>International Journal of Computer Applications</secondary-title></titles><periodical><full-title>International Journal of Computer Applications</full-title></periodical><pages>{45-50}</pages><volume>Volume 114 – Number 12</volume><dates><year>2015</year></dates><urls></urls><electronic-resource-num>10.5120/20034-2145</electronic-resource-num></record></Cite></EndNote>12.

2.5.1.1 Neural network approach with optimization techniques
In 2010, Shen et al. proposed a method which was a spiking neural network where the information is encoded at different spike of the EEG signal. Because of the limitation of the previous proposed methods in local minima and disability of coverage efficiency when negative synaptic weights were used a new method was proposed based on the partical awarm optimization (PSO). In this method the reported accuracy was 98.67% for training set and 97.33% for testing set.

After this, in 2015 Nesibe et al. developed a hybrid PSO combination and back-propagation algorithm for improved training of ANN. And in 2012 Seema et al. developed another hybrid method using ANN with PSO to detect epileptic seizure ADDIN EN.CITE <EndNote><Cite><Author>Saini</Author><Year>2017</Year><RecNum>2516</RecNum><DisplayText>13</DisplayText><record><rec-number>13</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459560″>13</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Saini, J.</author><author>Dutta, M.</author></authors></contributors><auth-address>a Department of Electronics and Communication Engineering , National Institute of Technical Teachers Training and Research , Chandigarh , India.</auth-address><titles><title>An extensive review on development of EEG-based computer-aided diagnosis systems for epilepsy detection</title><secondary-title>Network</secondary-title></titles><periodical><full-title>Network</full-title></periodical><pages>1-27</pages><volume>28</volume><number>1</number><edition>2017/05/26</edition><keywords><keyword>Diagnosis, Computer-Assisted/*methods</keyword><keyword>Electroencephalography/*methods</keyword><keyword>Epilepsy/*diagnosis/physiopathology</keyword><keyword>Humans</keyword><keyword>Artificial neural network (ANN)</keyword><keyword>electroencephalogram (EEG)</keyword><keyword>epilepsy</keyword><keyword>neural disorders</keyword><keyword>object recognition</keyword><keyword>wavelets</keyword></keywords><dates><year>2017</year></dates><isbn>1361-6536 (Electronic) 0954-898X (Linking)</isbn><accession-num>28537461</accession-num><urls><related-urls><url>https://www.ncbi.nlm.nih.gov/pubmed/28537461</url></related-urls></urls><electronic-resource-num>10.1080/0954898X.2017.1325527</electronic-resource-num></record></Cite></EndNote>13.

2.5.2 Wavelet Based
Though scientific work on seizure prediction and detection started around 1970, it took around 30 years to see the light of success. Among many successful methods, in time frequency methods the wavelet based seizure prediction and detection system stands out in terms of algorithmic elegance and efficiency. In EEG signal Wavelet transform (WT) captures the subtle changes as well PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5GYXVzdDwvQXV0aG9yPjxZZWFyPjIwMTU8L1llYXI+PFJl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ADDIN EN.CITE PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5GYXVzdDwvQXV0aG9yPjxZZWFyPjIwMTU8L1llYXI+PFJl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ADDIN EN.CITE.DATA 14. Raw EEG signals are suffer from poor low signal noise and artifacts. Wavelet based methods are well known for remove noise from the signals. Beside wavelet can be used for feature extraction.

Fiq: Examples of wavelets for EEG processing.

As there are no standard method adopted worldwide for studying and analysis of EEG signals so the performance parameters like accuracy, sensitivity, selectivity changes according to technique used. There are various techniques are evolved over couple of decades but the wavelet transform helps to improve understanding of EEG signals.

In 2012 Prof. Rashavan et.al has found that sensitivity of 99% is achieved with the help of wavelet transform as feature extraction and ANFIS as classifier but the only limitations of the systems is that the other performance parameters need to be improved.

Further research is carried out by Dr. T.R. Rangaswamy et.al ADDIN EN.CITE <EndNote><Cite><Author>D. Najumnissa</Author><Year>2012</Year><RecNum>2609</RecNum><DisplayText>15</DisplayText><record><rec-number>15</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459560″>15</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>D. Najumnissa, B. Pushpa and T.R. Rangaswamy</author></authors></contributors><titles><title>Detection and Classification of Seizures Using Feature Extrication and Particle Swarm Optimization Neural Network</title></titles><dates><year>2012</year></dates><urls></urls></record></Cite></EndNote>15 with AWT as Feature extraction method and ANFIS as classifier. He has come across the solution that the accuracy of the system is ranging from 97%-99% but there should be work done to improve selectivity and sensitivity. Miss. Ashwini Bhople et.al has compared the various different techniques used in study and analysis of epileptic seizure detection. She came with the finding that the accuracy depends on the methodology used in system.

L. M. patnaik et.al has used WT and ANN for analysis of EEG signals and he has found that the specificity is 99.12% and sensitivity of 91.29% is achieve. Hasan Ocak et.al ADDIN EN.CITE <EndNote><Cite><Author>Ocak</Author><Year>2009</Year><RecNum>2611</RecNum><DisplayText>16</DisplayText><record><rec-number>16</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459560″>16</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Ocak, Hasan</author></authors></contributors><titles><title>Automatic detection of epileptic seizures in EEG using discrete wavelet transform and approximate entropy</title><secondary-title>Expert Systems with Applications</secondary-title></titles><periodical><full-title>Expert Systems with Applications</full-title></periodical><pages>2027-2036</pages><volume>36</volume><number>2</number><section>2027</section><dates><year>2009</year></dates><isbn>09574174</isbn><urls></urls><electronic-resource-num>10.1016/j.eswa.2007.12.065</electronic-resource-num></record></Cite></EndNote>16uses different approach he used approximate entropy and WT for feature extraction and he found accuracy has increased to 96% but having problem of selecting appropriate wavelet functions. Subasi et.al ADDIN EN.CITE <EndNote><Cite><Author>Subasi</Author><Year>2007</Year><RecNum>2612</RecNum><DisplayText>17</DisplayText><record><rec-number>17</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>17</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Subasi, A.</author></authors></contributors><titles><title>EEG signal classification using wavelet feature extraction and a mixture of expert model</title><secondary-title>Expert Systems with Applications</secondary-title></titles><periodical><full-title>Expert Systems with Applications</full-title></periodical><pages>1084-1093</pages><volume>32</volume><number>4</number><section>1084</section><dates><year>2007</year></dates><isbn>09574174</isbn><urls></urls><electronic-resource-num>10.1016/j.eswa.2006.02.005</electronic-resource-num></record></Cite></EndNote>17 has used combination of WT and ANFIS and found specificity of 93.7% and sensitivity of 94.3% which is less as compared with work done by Dr. T. R. Rangaswamy though both used same techniques. Aarabi et.al ADDIN EN.CITE <EndNote><Cite><Author>Aarabi</Author><Year>2009</Year><RecNum>2613</RecNum><DisplayText>18</DisplayText><record><rec-number>18</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>18</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Aarabi, A.</author><author>Fazel-Rezai, R.</author><author>Aghakhani, Y.</author></authors></contributors><auth-address>Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, MB, Canada. [email protected]</auth-address><titles><title>A fuzzy rule-based system for epileptic seizure detection in intracranial EEG</title><secondary-title>Clin Neurophysiol</secondary-title></titles><periodical><full-title>Clin Neurophysiol</full-title></periodical><pages>1648-57</pages><volume>120</volume><number>9</number><edition>2009/07/28</edition><keywords><keyword>Adolescent</keyword><keyword>Adult</keyword><keyword>Algorithms</keyword><keyword>Data Interpretation, Statistical</keyword><keyword>Electrodes, Implanted</keyword><keyword>*Electroencephalography</keyword><keyword>Entropy</keyword><keyword>Epilepsy/*diagnosis</keyword><keyword>False Positive Reactions</keyword><keyword>Female</keyword><keyword>Fuzzy Logic</keyword><keyword>Humans</keyword><keyword>Male</keyword><keyword>Middle Aged</keyword><keyword>Seizures/*diagnosis</keyword><keyword>Young Adult</keyword></keywords><dates><year>2009</year><pub-dates><date>Sep</date></pub-dates></dates><isbn>1872-8952 (Electronic) 1388-2457 (Linking)</isbn><accession-num>19632891</accession-num><urls><related-urls><url>https://www.ncbi.nlm.nih.gov/pubmed/19632891</url></related-urls></urls><electronic-resource-num>10.1016/j.clinph.2009.07.002</electronic-resource-num></record></Cite></EndNote>18 has found that the understanding of EEG signals are better if used wavelet transform and one can easily develop computational model for better understanding of EEG signals. Inan Guler et.al ADDIN EN.CITE <EndNote><Cite><Author>Guler</Author><Year>2005</Year><RecNum>2614</RecNum><DisplayText>19</DisplayText><record><rec-number>19</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>19</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Guler, I.</author><author>Ubeyli, E. D.</author></authors></contributors><auth-address>Department of Electronics and Computer Education, Faculty of Technical Education, Gazi University, 06500 Teknikokullar, Ankara, Turkey. [email protected]</auth-address><titles><title>Adaptive neuro-fuzzy inference system for classification of EEG signals using wavelet coefficients</title><secondary-title>J Neurosci Methods</secondary-title></titles><periodical><full-title>J Neurosci Methods</full-title></periodical><pages>113-21</pages><volume>148</volume><number>2</number><edition>2005/08/02</edition><keywords><keyword>Artificial Intelligence</keyword><keyword>Brain/*physiology</keyword><keyword>Electroencephalography/*methods</keyword><keyword>Evoked Potentials/*physiology</keyword><keyword>*Fuzzy Logic</keyword><keyword>Humans</keyword><keyword>Neural Networks (Computer)</keyword><keyword>*Signal Processing, Computer-Assisted</keyword><keyword>Software</keyword></keywords><dates><year>2005</year><pub-dates><date>Oct 30</date></pub-dates></dates><isbn>0165-0270 (Print) 0165-0270 (Linking)</isbn><accession-num>16054702</accession-num><urls><related-urls><url>https://www.ncbi.nlm.nih.gov/pubmed/16054702</url></related-urls></urls><electronic-resource-num>10.1016/j.jneumeth.2005.04.013</electronic-resource-num></record></Cite></EndNote>19 found that if the numbers of stages are increased then the more accuracy can be achieved and using WT and ANFIS he got classification accuracy as 98.68%. From the above literature survey it has found that wavelet transform helps to improve understanding of EEG signals as well as performance parameters of the mathematical results.

2.5.3 Furrier Transformation Based
Prior et al. (1973) provided the first solution to the problem of automated epilepsy detection. During their studies on intensive therapy unit where patients were monitored for acute anoxic episodes, it was observed that an automated device used for monitoring EEG activity of brain signals was of great importance. It was able to recognize the presence of seizure discharges as well as it analyzed the actual movements of patients during seizure activity. The researchers realized that it is possible to monitor seizure activities on the basis of amplitude changes. They used signal-processing algorithms and applied Fourier transform for removal of the artifact. Meenakshi et al. (2014) developed the analysis of frequencies for epileptic seizures and healthy EEG signals by using fast Fourier transform technique. They divided recorded EEG signal activities into five different frequency domains: ?, ?, ?, ?, and ? sets. The respective frequency distribution over FFT was compared to analyze differences between healthy and epileptic signals.

2.5.4. Classifier based techniques
Saadat and Hossein (2013) proposed an epileptic seizure detection algorithm with patient-specific features. The nonseizure- and seizure-oriented EEG signals were applied to the system and then discrete Fourier transform and DWT were applied on the following five special frequency bands: ?, ?, ?, ?, and ?. After obtaining feature-specific maximum distinction between both classes, the PSO-based classifier was used to determine the optimal linear decision boundary. By using this classifier, it is possible to make adjustments for parameters such as latency, specificity, and sensitivity. As a result, smaller latency with higher sensitivity range was obtained with application of proposed algorithm. Out of 161 seizure inputs, it was able to recognize 157 seizures accurately and average sensitivity with above system performance was reported to be 98%.
Sivasanskari et al. (2013) presented an advanced approach for automated detection of epileptic seizures using multilayer perceptron neural network type classifier. A statistical tool named as independent component analysis (ICA) was used for extraction of features. With the generation of fast, memory efficient and scaled conjugate gradient back-propagation algorithm, all the ascertained signals were trained. The algorithm was designed using tan-sigmoid transfer function over output layer as well as hidden layer and its performance on specific data set was realized using confusion matrix and mean square error function. The detection accuracy was observed to be 100% with the proposed system.

Yaozhang et al. (2007) developed a support vector machine (SVM)-based design for sharp epileptic signals in EEG spectrum. This technique helps to map intracranial EEG time series into some corresponding novelty sequence that was classified on the basis of energy-based statistics over a short duration of the window. A comparison with numerical studies between SVM and simple ANN approach was derived that proves SVM algorithm as a better tool for epileptic seizure detection. This classifier-based technique provided 100% accuracy.

Dhande and Gulhane (2015) used two different stages: the first stage feature extraction was performed using few statistical measures and in the second phase ANN models with back propagation algorithm-based classifiers were tested. It resulted in higher training performance along with improved classification accuracy up to 100%.

Sharmila and Geethanjali (2016) applied novel pattern recognition methods for detection of epilepsy-related abnormalities in EEG signals. The statistical features of the signal were derived using DWT coefficients ranging between D3 and D5, whereas for seizure abnormality detection A5 was used along with naïve Bayes. The K-nearest neighbor classifier was proven to be useful for 14 unique combinations of sets A–D and E. The applied feature selection and ranking-based technique for mutual information estimation was proven to be helpful in epilepsy detection and it improved accuracy up to 100%.

Asha et al. (2013) used combination of different features of EEG signals for multichannel detection and these uniqueness parameters were detected using a 4-s window that creates a feature vector. ANN/SVM classifier is trained via feature vectors and a carefully selected data set so that the rule set can help in removal of interictal spikes and proper detection of seizures. With proposed system researchers were able to record overall accuracy of 75%.

Mousavi et al. (2008) presented the latest method for detection of epilepsy with autoregressive estimation. The optimum order of AR model is calculated using Bayesian information criterion (BIC) and different EEG parameters along with sub-band details are analyzed. These obtained parameter analyses are used for classification of EEG signals along with multilayer perceptron classifier and accuracy for classification was observed to be 96%.

Minfen et al. (2009) proposed a local SVM-type method for modeling of EEG signals. By using the local method, the speed of EEG prediction can be increased and the proposed method was also capable enough to detect EEG signals that follow dynamical characteristics with the difference in epileptic and normal signal waveforms. The local SVM approach effectively improves detection precision and prediction.

Pega et al. (2003) proposed a method to detect epileptic seizures in newborn babies with EEG signals. The scheme is actually based upon observation of changing the behavior of wavelet coefficients of EEG signals on various scales. Mutual information evaluation function (MIEF) was used to get information about optimal features of signals and a subset of specific features was obtained after analysis. The subset is then applied as an input to ANN classifier for organizing EEG signal into the set of nonseizure and seizure activity. They reported seizure detection rate of 96.35% and false alarm rate of 6.2% for this research study.

Nasser et al. (2006) worked on the comparison of different classifiers for detection and control of Epilepsy seizures. The classifier applied on normal and epileptic signals was adaptive neural fuzzy network (ANFIS) and its performance was compared with some previously tested classifiers such as feed-forward back-propagation neural network (FBNN), ANFIS, SVM, etc. This detection technique was able to reach up to an accuracy level of 85.9%.

Saadat and Hossein (2011) observed that if classification and feature extraction algorithms are applied carefully, then successful epileptic seizure detection can be performed. Hence, they proposed an onset detection method that works on the basis of dynamic cascade feed-forward neural networks (DCFNNs). First of all, spatial and spectral features of non-seizure and L-second seizure activity were extracted and then DCFNN algorithm was applied. The two major advantages observed with this algorithm were as follows: it was possible to create the maximum distinction between both classes using extracted features. Second, DCFNN-type classifier uses to have parallel structure, and hence it converges to form an optimal classifier whenever the size of representative training sets increases. Ultimately, this method resulted into smaller latency value of 55% with higher improvement in sensitivity with practical value of 98%.

Harikumar et al. (2016) evaluated performance of PSO and SVM for robust classification of Epilepsy disease. The input EEG signals were first sampled and then their artifacts were removed in order to deal with the problem of higher dimensionality, the PSD procedures were applied. Finally, the data set is processed with SVM along with its kernels and the post classification work is performed using PSO. The performance comparison is done to get useful results for epilepsy classification. Final results for this study were presented in terms of high performance index and quality value with practical value of 98.24% and 24.19% only.

Samanwoy et al. (2008) presented a neural network-based classifier with latest principal component analysis (PCA) type enhanced cosine radial basis function. Classification between interictal, ictal, and healthy EEG signals was done on the basis of the mixed band wavelet chaos technology that was integrated with a two-stage classifier. The input stage of both classifiers consisted of nine parameters mixed band-type feature space. PCA methodology was applied at the very first stage for accurate feature enhancement. The RBFNN is applied to the second stage of the system so that classification accuracy can be improved with perfect rearrangement of input space as per principal signal components. With the sensitivity and extensive parametric analysis, authors performed the validation task for accuracy as well as the robustness of classifier. An EEG classification accuracy of 96.6% was obtained with wavelet chaos-type neural network system and it was observed that system was robust enough to all changes that occurred in training data whereas standard deviation was recorded to be 1.4%. In the case of epilepsy diagnosis, the classification accuracy reached up to 99.3% for interictal and normal EEG signals.

Akshata et al. (2016) used wavelet-based neural network classifier to recognize the EEG signal disturbances. The basic technique involved separation of recorded EEG signal into different frequency bands in terms of ?, ?, ?, ?, and ? frequencies. It was done using multiresolution analysis and DWT method. For detection of seizure onset, the wavelet analysis techniques were used and further filtered signal output was used for calculation of spectral power ratio.

The recognized features were processed using ANN classifier and it helped to detect the seizure and non-seizure activities from human brain with accuracy equal to 96%.

Shail and Rao (2014) proposed a model for analyzing seizure and non-seizure EEG signals. The method used for extraction of IMFs is empirical mode decomposition and based on the mean weighted frequency various features are extracted; finally, the extracted features are classified using neural network. The proposed techniques of this paper are used for processing of EEG signals so that an advanced tool can be developed for physicians to diagnose abnormalities in brain functioning. This method has potential to provide lots of benefits for the diagnosis like specificity, good sensitivity, high accuracy, and fast diagnosis. Reported average accuracy was 99.8% with the above proposed methodology.

Aguirre-Echeverry et al. (2013) proposed a technique based on one-class classifiers that are preferred due to their high performance even under balanced classes as well as in lack of provided target data in the form of bio-signals. Several aspects taken under considerations include rejection rate parameter and kernel parameter that are related to performance stability and computational cost. The proposed work was able to provide improved performance using automatic tuning algorithms. A mixture of Gaussian and support vector data descriptor is used to get higher stability and better convergence time. With this method, specificity and sensitivity were improved up to 100% and 96.67%, respectively.

Orellana and Fabio (2016) presented an advanced computational method for detection of epilepsy disease. The proposed technique includes an offline patient-dependent system that works on the basis of random forest classifier to detect the seizures in brainwaves. Features from brainwaves are extracted using one-dimension information that is generated from the spectro-temporal transformation of bio-signals when they pass through an envelope detector. They recorded results for false positive rate, sensitivity, and specificity and the recorded values were 0.77 h?1, 99.29%, and 97.12%, respectively.

Meenakshi et al. (2014) presented a computer-aided diagnostic system for EEG data so that classification of epileptic and nonepileptic brain can be performed. The proposed methods can be used over hardware monitoring systems and the techniques used for classification were ANN, SVM, K-means classifier, naïve Bayes classifier, and radial basis function. They presented average accuracy as well as specificity of 100% for classification of epileptic signals from normal EEG data.

Hiram et al. (2007) designed a patient-specific detector for epilepsy seizures with genetic programming algorithm for artificial features. This general purpose program consists of methodic algorithm made up of a genetic module for programming and k-nearest neighbor-type classifier for the creation of synthetic features. In this paper, the artificial features of EEG signals were constructed with the reconstruction of state space trajectories so that conditions of elliptic seizures can be classified. The analyses were able to set a benchmark for comparison between healthy and unhealthy brain signals as 88 out of total 92 seizures were detected accurately and the calculated low false negative rates were about 4.35%.

Kemal and Salih (2007) developed a hybrid system for detection of epileptic seizures from EEG-based data using the combination of fast Fourier transform and decision tree classifier. The first phase of the study was feature extraction stage that is based upon FFT method, whereas the second stage works for decision-making processes by following decision tree classifier. The final validation of this analysis was completed using k-fold cross-validation process and practical values for specificity, sensitivity, and accuracy were observed. With five-fold cross validation, the obtained accuracy was 98.68% whereas with 10 fold cross-validation the results went up to 98.72% in terms of accuracy.

Patnaik and Ohil (2008) proposed a wavelet-based classic feature extraction technique with some simple statistical parameters that can easily detect Epileptic activities using back propagation neural network as a classifier. A post-classification stage is also used in this mechanism that can effectively correlate outputs of various channels with increased accuracy, sensitivity, and specificity having practical values of 91.14%, 91.29% and 99.19%, respectively.

Kiranmayi and Udayashankara (2013) presented a bispectrum analysis system for EEG signal-based epilepsy detection; this system is able to characterize all nonlinearities of the signal with ease. The features that are extracted from EEG bispectrum are further applied to neural network classifier for detection of signals whether they are epileptic or normal. The observed results were far better than conventional power spectrum analysis with an accuracy of about 81.67%.

2.5.5 Other techniques
Karthika and Vijayanand (2016) adopted the android platform for fast processing of EEG signals on the real-time basis. It helps in heart rate variability (HRV) analysis as well as in arrhythmia classification. The EEG data were first acquired and transmitted to the android terminal on a smart phone using Bluetooth technology. A signal-conditioning unit was connected with EEG simulator and then received an analog version of signals was converted to digital form and they were further passed to a microcontroller. This device assists in signal transfers to mobile phones using Bluetooth and an android-based application serve with complete signal-monitoring services. It helped to generate real-time regular updates about EEG signals and seizure spikes and overall accuracy of proposed system was reported to be 96%.

Dalia et al. (2015) presented an integrated android platform for advanced detection of seizure activities. It also serves as a location access provider while tracking and monitoring patients with basic first aid information. The developed system used to monitor patient’s movements and whenever a seizure activity is observed than an immediate message is transferred to registered caregivers along with personals location power by global positioning system (GPS) system. At the same time, this android application used to provide first aid details on the mobile screen so that patient’s health can be recovered soon.

Gotman and Gloor (1976) made an attempt to quantify and recognize interictal epileptic activity that is in the form of sharp waves and spikes using a small computer. In order to perform this automatic recognition procedure, the EEG waveform of every channel was broken down into the half waves.

Few advanced procedures were applied for the elimination of unwanted sharp or spike-like waveforms that were generated by muscle potentials, eye blinks, and sharp ?-activities. This system presented a potential solution for clinical applications of EEG-based epilepsy detection.

Thomas et al. (1974) described a simple analog-digital circuit that can accept one or more EEG signal inputs and as a result, it gives information about the presence of seizure activity. This paper demonstrated the telemetered EEG monitoring for reliable and routine seizure detection.

Ping et al. (2012) analyzed that the basic time–frequency features of EEG signals can be easily extracted using sparse representation using Marching Pursuit Algorithm. Although the computation burden becomes so high if we apply MP in the real sense but in order to reduce complexity in calculations of sparse representation, adopt harmony type search method is used that finds best atoms from search space. Using this approach on the basis of time–frequency dictionary, it was possible to enhance the performance of epileptic seizure classification with maximum accuracy of 100%.

William et al. (1978) took a two-dimensional-type amplitude duration sample space where epileptic transients (ETs) of EEG signal that are differentiated in terms of duration and amplitude were compared to the background activity using a cluster analysis. An optimum boundary value was assigned to clusters and then appropriate decisions about ET and background were carried out. The recorded EEG signals were filtered by applying linear prediction and second differentiation techniques so that desired degree of separation can be achieved. It helped to enhance the probability of accurate decisions by the great value.

Lopes Da Silva et al. (1977) presented a new model for automatic EEG analysis with automatic non-stationary detection (ASD) that is based on autoregressive filet model and the concept of inverse filtering. Using this technique, it was possible to identify the presence of transient nonstationarities for inter-ictal EEG. This ASD technique was implemented using a general-purpose-type digital computer and it was able to perform multichannel analysis as well as the statistical assessment of paroxysmal patterns ADDIN EN.CITE <EndNote><Cite><Author>Lopes da Silva</Author><Year>1977</Year><RecNum>2635</RecNum><record><rec-number>20</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>20</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Lopes da Silva, F. H.</author><author>Van Hulten, K.</author><author>Lommen, J. G.</author><author>Storm Van Leeuwen, W.</author><author>Van Veelen, C. W. M.</author><author>Vliegenthart, W.</author></authors></contributors><titles><title>Automatic detection and localization of epileptic foci</title><secondary-title>Electroencephalography and Clinical Neurophysiology</secondary-title></titles><periodical><full-title>Electroencephalography and Clinical Neurophysiology</full-title></periodical><pages>1-13</pages><volume>43</volume><number>1</number><dates><year>1977</year><pub-dates><date>1977/07/01/</date></pub-dates></dates><isbn>0013-4694</isbn><urls><related-urls><url>http://www.sciencedirect.com/science/article/pii/0013469477901894</url></related-urls></urls><electronic-resource-num>https://doi.org/10.1016/0013-4694(77)90189-4</electronic-resource-num></record></Cite></EndNote>{Lopes da Silva, 1977 #2635}.

Ktonas et al. (1979) proposed a simple detection algorithm for artifacts that can be applied on the large quality of EEG data ADDIN EN.CITE <EndNote><Cite><Author>Ktonas</Author><Year>1979</Year><RecNum>2630</RecNum><record><rec-number>21</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>21</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Ktonas, P. Y.</author><author>Osorio, P. L.</author><author>Everett, R. L.</author></authors></contributors><titles><title>Automated detection of EEG artifacts during sleep: Preprocessing for all-night spectral analysis</title><secondary-title>Electroencephalography and Clinical Neurophysiology</secondary-title></titles><periodical><full-title>Electroencephalography and Clinical Neurophysiology</full-title></periodical><pages>382-388</pages><volume>46</volume><number>4</number><dates><year>1979</year><pub-dates><date>1979/04/01/</date></pub-dates></dates><isbn>0013-4694</isbn><urls><related-urls><url>http://www.sciencedirect.com/science/article/pii/0013469479901391</url></related-urls></urls><electronic-resource-num>https://doi.org/10.1016/0013-4694(79)90139-1</electronic-resource-num></record></Cite></EndNote>{Ktonas, 1979 #2630}.It was useful to apply spectral analysis techniques over a general-purpose computer system when visual inspection of large data becomes difficult{Ktonas, 1979 #2630}{Ktonas, 1979 #2630}{, #[email protected]@hidden}{, #2629}{Ktonas, 1979 #2631}{Ktonas, 1979 #2630}{Ktonas, 1979 #2630}{Ktonas, 1979 #2630}. A technique was developed using chi-square goodness test over Gaussian distribution and EEG epochs of almost 30-s duration were analyzed using this method. The test produced a very large value of chi-square variable whenever artifacts were present in the signal and hence the signal non-stationarities were used perfectly for analysis of the presence of epileptic seizures.

Gotman et al. (1979) recorded EEG signal from sphenoid and scalp electrodes using a telemetry system and PDP12 computer. This method was able to analyze EEG signal over 16 channels at a time and detected spikes after every 1 s were saved to magnetic tape for records. After complete monitoring, this tape was simply played back using EEG machine and all the discontinuous spike sections were easily observed. It was further sent to a computer system for displaying and determining temporal and spatial distributions of epileptic waves. Accuracy for this study was reported in the range of 67%–100% ADDIN EN.CITE <EndNote><Cite><Author>Gotman</Author><Year>1979</Year><RecNum>2628</RecNum><DisplayText>20</DisplayText><record><rec-number>22</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>22</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Gotman, J.</author><author>Ives, J. R.</author><author>Gloor, P.</author></authors></contributors><titles><title>Automatic recognition of inter-ictal epileptic activity in prolonged EEG recordings</title><secondary-title>Electroencephalogr Clin Neurophysiol</secondary-title><alt-title>Electroencephalography and clinical neurophysiology</alt-title></titles><alt-periodical><full-title>Electroencephalography and Clinical Neurophysiology</full-title></alt-periodical><pages>510-20</pages><volume>46</volume><number>5</number><edition>1979/05/01</edition><keywords><keyword>*Electroencephalography</keyword><keyword>Epilepsy/physiopathology</keyword><keyword>Humans</keyword><keyword>*Pattern Recognition, Automated</keyword></keywords><dates><year>1979</year><pub-dates><date>May</date></pub-dates></dates><isbn>0013-4694 (Print) 0013-4694</isbn><accession-num>88339</accession-num><urls></urls><remote-database-provider>NLM</remote-database-provider><language>eng</language></record></Cite></EndNote>20.
Gotman (1982) developed the automatic detection method for EEG signal of seizures spikes. First of all, EEG is decomposed into the elementary waves and then simple procedures were implemented to get information about amplitude, duration, and rhythmicity of waves. This method was tested over 24 surface recordings as well as on 44 recordings of intracerebral electrodes and results showed an increased system performance for real-time seizure spike recognition ADDIN EN.CITE <EndNote><Cite><Author>Gotman</Author><Year>1979</Year><RecNum>2628</RecNum><DisplayText>20</DisplayText><record><rec-number>22</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459561″>22</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Gotman, J.</author><author>Ives, J. R.</author><author>Gloor, P.</author></authors></contributors><titles><title>Automatic recognition of inter-ictal epileptic activity in prolonged EEG recordings</title><secondary-title>Electroencephalogr Clin Neurophysiol</secondary-title><alt-title>Electroencephalography and clinical neurophysiology</alt-title></titles><alt-periodical><full-title>Electroencephalography and Clinical Neurophysiology</full-title></alt-periodical><pages>510-20</pages><volume>46</volume><number>5</number><edition>1979/05/01</edition><keywords><keyword>*Electroencephalography</keyword><keyword>Epilepsy/physiopathology</keyword><keyword>Humans</keyword><keyword>*Pattern Recognition, Automated</keyword></keywords><dates><year>1979</year><pub-dates><date>May</date></pub-dates></dates><isbn>0013-4694 (Print) 0013-4694</isbn><accession-num>88339</accession-num><urls></urls><remote-database-provider>NLM</remote-database-provider><language>eng</language></record></Cite></EndNote>20.

Richard and Swartz (1989) used electroconvulsive therapy (ECT) for automatic monitoring of seizure activity and specific EEG parameter like integrated voltage was calculated. This detected voltage signal was further converted into a digital sequence compared with the threshold reference values of some standard signal ADDIN EN.CITE <EndNote><Cite><Author>Abrams</Author><Year>1989</Year><RecNum>2626</RecNum><DisplayText>21</DisplayText><record><rec-number>23</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459562″>23</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Richard S. Abrams</author><author>Conrad M. Swartz</author></authors></contributors><titles><title>ELECTROCONVULSIVE THERAPY APPARATUS AND METHOD FOR MONITORNG PATENT SEZURES</title><secondary-title>US Patent</secondary-title></titles><periodical><full-title>US Patent</full-title></periodical><number>48 78 498</number><dates><year>1989</year><pub-dates><date>October 17</date></pub-dates></dates><urls></urls></record></Cite></EndNote>21. This comparison provided information about seizure occurrence via moving strip chart paper, oscilloscope Cathode Ray Tube (CRT) screen, or alphanumerical elapsed time display system.

Harding (1993) developed an automated monitoring system for recording patient electrode activities during intractable seizures. The system worked upon a flowing graphic image generated from a 32- channel real-time electrocorticographic (ECoG) network, a mechanism for automated detection of seizures, and a 6-min recording set of ECoG before the actual occurrence of a seizure. When tests were applied over 40 patients, then this automated system was observed to present recording and seizure detection accuracy of about 95% for clinical analysis and 86% accuracy rate was calculated for patient-specific patterns of seizure activities ADDIN EN.CITE <EndNote><Cite><Author>Harding</Author><Year>1993</Year><RecNum>2625</RecNum><DisplayText>22</DisplayText><record><rec-number>24</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459562″>24</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Harding, G. W.</author></authors></contributors><titles><title>An automated seizure monitoring system for patients with indwelling recording electrodes</title><secondary-title>Electroencephalography and Clinical Neurophysiology</secondary-title></titles><periodical><full-title>Electroencephalography and Clinical Neurophysiology</full-title></periodical><pages>428-437</pages><volume>86</volume><number>6</number><keywords><keyword>Seizure monitor</keyword><keyword>Automatic seizure detection</keyword><keyword>Indwelling electrodes</keyword></keywords><dates><year>1993</year><pub-dates><date>1993/06/01/</date></pub-dates></dates><isbn>0013-4694</isbn><urls><related-urls><url>http://www.sciencedirect.com/science/article/pii/001346949390138L</url></related-urls></urls><electronic-resource-num>https://doi.org/10.1016/0013-4694(93)90138-L</electronic-resource-num></record></Cite></EndNote>22.

Dale et al. (1994) developed a patient-monitoring system so that seizures can be automatically detected by observing electrical discharge signals from the brain. The electrical discharge signal developed in the brain was converted into a digital sequence and then send as an input to a microprocessor system ADDIN EN.CITE <EndNote><Cite><Author>OE</Author><Year>1994</Year><RecNum>2623</RecNum><DisplayText>23</DisplayText><record><rec-number>25</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459562″>25</key></foreign-keys><ref-type name=”Journal Article”>17</ref-type><contributors><authors><author>Dale OE</author><author>Ronald LP</author><author>John HC</author><author>Webber WRS</author><author>John CA</author></authors></contributors><titles><title>AUTOMATIC DETECTION OF SEZURES USING ELECTROENCEPHALOGRAPHC SIGNALS</title><secondary-title>US Patent</secondary-title></titles><periodical><full-title>US Patent</full-title></periodical><dates><year>1994</year><pub-dates><date>May 17</date></pub-dates></dates><urls></urls><custom7>5311876</custom7></record></Cite></EndNote>23. The microprocessor system further detects seizure activities by simply dividing the digital sequence into different time segments; these time segments were pre-processed with standardized signals and then specific features were extracted to collect information about seizure occurrence.

Kristin et al. (2001) presented an effective alternative for irreversible, destructive, and respective surgical treatment using interictal spike patterns along with continuous EEG signal measurement so that the seizure activity can be controlled using electrical and chemical control systems PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5KZXJnZXI8L0F1dGhvcj48WWVhcj4yMDAxPC9ZZWFyPjxS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==
ADDIN EN.CITE PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5KZXJnZXI8L0F1dGhvcj48WWVhcj4yMDAxPC9ZZWFyPjxS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==
ADDIN EN.CITE.DATA 24. Due to the complete insensitivity to waveform amplitude changes, this method was least affected by noise and hence it delivered the most robust performance as compared with other existing techniques.

Leon (2003) presented an overview of signal-processing methodologies for analyses of epileptic seizures with its nonlinear dynamics model. The analysis resulted in useful information for the clinical and medical industry for getting control over underlying disorders ADDIN EN.CITE <EndNote><Cite><Author>Iasemidis</Author><Year>2003</Year><RecNum>2620</RecNum><DisplayText>25</DisplayText><record><rec-number>27</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459562″>27</key></foreign-keys><ref-type name=”Book”>6</ref-type><contributors><authors><author>Leon D. Iasemidis</author></authors></contributors><titles><title>Epileptic Seizure Prediction and Control</title></titles><volume>50</volume><num-vols>5</num-vols><dates><year>2003</year><pub-dates><date>May</date></pub-dates></dates><publisher>IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING</publisher><urls></urls></record></Cite></EndNote>25. The significance of this research study shows that it is important to develop a prediction/warning algorithm that can either work as software modules between the compact monitoring device and desktop or present a miniaturized hardware-type solution that can be implanted directly on the body of the patient.

Ralph et al. (2008) presented a novel procedure for online, generic, and real-time automated detection while working on multi-morphologic ictal patterns of EEG waveforms. The following are the six common ictal morphology types: poly spikes, amplitude depression, ?-rhythmic activity, ?, ?, and ?. It is observed that the detection performance of a system can be increased by taking morphology into account, and in proposed method the detection rate was increased up to 96% PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5NZWllcjwvQXV0aG9yPjxZZWFyPjIwMDg8L1llYXI+PFJl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ADDIN EN.CITE PEVuZE5vdGU+PENpdGU+PEF1dGhvcj5NZWllcjwvQXV0aG9yPjxZZWFyPjIwMDg8L1llYXI+PFJl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ADDIN EN.CITE.DATA 26.

Sabrina et al. (2016) proposed an advanced framework for epilepsy detection using fast potential hierarchical agglomerative (PHA) clustering methods along with empirical mode decomposition. The presented algorithm was capable enough to count IMFs of input EEG signals and further it works on Kolmogorov distance among all IMFs. Finally, the detection was performed on the basis of PHA method. The evaluation resulted in the accuracy rate of 98.84% ADDIN EN.CITE <EndNote><Cite><Author>Belhadj</Author><Year>2016</Year><RecNum>2617</RecNum><DisplayText>27</DisplayText><record><rec-number>29</rec-number><foreign-keys><key app=”EN” db-id=”asa9pf9ebz5df8e0tw75zafcda2arpdt5ptx” timestamp=”1531459563″>29</key></foreign-keys><ref-type name=”Book”>6</ref-type><contributors><authors><author>Sabrina Belhadj</author><author>Abedlouaheb Attia†</author><author>Bachir Ahmed Adnane</author><author>Zoubir Ahmed-Foitih </author><author>Abdelmalik Taleb Ahme</author></authors></contributors><titles><title>A Novel Epileptic Seizure Detection Using Fast Potential-based Hierarchical Agglomerative Clustering Based on EMD</title></titles><volume>16</volume><num-vols>5</num-vols><dates><year>2016</year><pub-dates><date>May</date></pub-dates></dates><pub-location>IJCSNS International Journal of Computer Science and Network Security</pub-location><urls></urls></record></Cite></EndNote>27.

ADDIN EN.REFLIST 1A. Jacoby, D. Snape, and G. A. Baker, “Epilepsy and social identity: the stigma of a chronic neurological disorder,” The Lancet Neurology, vol. 4, no. 3, pp. 171-178, 2005.

2H. W. S. Valerie Jewells, “Review of Epilepsy – Etiology, Diagnostic Evaluation and Treatment,” International Journal of Neurorehabilitation, vol. 01, no. 03, 2014.

3M. Angalakuditi, and N. Angalakuditi, “A comprehensive review of the literature on epilepsy in selected countries in emerging markets,” Neuropsychiatr Dis Treat, vol. 7, pp. 585-97, 2011.

4J. Kaur, and A. Kaur, “A review on analysis of EEG signals.” pp. 957-960.

5A. Mitra, A. Kraft, P. Wright, B. Acland, A. Z. Snyder, Z. Rosenthal, L. Czerniewski, A. Bauer, L. Snyder, J. Culver, J. M. Lee, and M. E. Raichle, “Spontaneous Infra-slow Brain Activity Has Unique Spatiotemporal Dynamics and Laminar Structure,” Neuron, vol. 98, no. 2, pp. 297-305 e6, Apr 18, 2018.

6B. S. Zainuddin, Z. Hussain, and I. S. Isa, “Alpha and beta EEG brainwave signal classification technique: A conceptual study.” pp. 233-237.

7A. E. Elisha, L. Garg, O. Falzon, and G. D. Giovanni, “EEG feature extraction using common spatial pattern with spectral graph decomposition.” pp. 1-8.

8M. Ahmad, M. Saeed, S. Saleem, and A. Kamboh, Seizure Detection using EEG:A survey of different Techniques, 2016.

9W. Weng, and K. Khorasani, “An Adaptive Structure Neural Networks with Application to EEG Automatic Seizure Detection,” Neural Networks, vol. 9, no. 7, pp. 1223-1240, 1996/10/01/, 1996.

10A. Subasi, “Epileptic seizure detection using dynamic wavelet network,” Expert Syst. Appl., vol. 29, no. 2, pp. 343-355, 2005.

11A. T. Tzallas, M. G. Tsipouras, and D. I. Fotiadis, “Epileptic seizure detection in EEGs using time-frequency analysis,” IEEE Trans Inf Technol Biomed, vol. 13, no. 5, pp. 703-10, Sep, 2009.

12G. D. Ç. Ö. Ç. M. R. Bozkurt, “The Detection of Normal and Epileptic EEG Signals using ANN Methods with Matlab-based GUI,” International Journal of Computer Applications, vol. Volume 114 – Number 12, pp. {45-50}, 2015.

13J. Saini, and M. Dutta, “An extensive review on development of EEG-based computer-aided diagnosis systems for epilepsy detection,” Network, vol. 28, no. 1, pp. 1-27, 2017.

14O. Faust, U. R. Acharya, H. Adeli, and A. Adeli, “Wavelet-based EEG processing for computer-aided seizure detection and epilepsy diagnosis,” Seizure, vol. 26, pp. 56-64, Mar, 2015.

15B. P. a. T. R. R. D. Najumnissa, “Detection and Classification of Seizures Using Feature Extrication and Particle Swarm Optimization Neural Network,” 2012.

16H. Ocak, “Automatic detection of epileptic seizures in EEG using discrete wavelet transform and approximate entropy,” Expert Systems with Applications, vol. 36, no. 2, pp. 2027-2036, 2009.

17A. Subasi, “EEG signal classification using wavelet feature extraction and a mixture of expert model,” Expert Systems with Applications, vol. 32, no. 4, pp. 1084-1093, 2007.

18A. Aarabi, R. Fazel-Rezai, and Y. Aghakhani, “A fuzzy rule-based system for epileptic seizure detection in intracranial EEG,” Clin Neurophysiol, vol. 120, no. 9, pp. 1648-57, Sep, 2009.

19I. Guler, and E. D. Ubeyli, “Adaptive neuro-fuzzy inference system for classification of EEG signals using wavelet coefficients,” J Neurosci Methods, vol. 148, no. 2, pp. 113-21, Oct 30, 2005.

20J. Gotman, J. R. Ives, and P. Gloor, “Automatic recognition of inter-ictal epileptic activity in prolonged EEG recordings,” Electroencephalogr Clin Neurophysiol, vol. 46, no. 5, pp. 510-20, May, 1979.

21R. S. Abrams, and C. M. Swartz, “ELECTROCONVULSIVE THERAPY APPARATUS AND METHOD FOR MONITORNG PATENT SEZURES,” US Patent, no. 48 78 498, October 17, 1989.

22G. W. Harding, “An automated seizure monitoring system for patients with indwelling recording electrodes,” Electroencephalography and Clinical Neurophysiology, vol. 86, no. 6, pp. 428-437, 1993/06/01/, 1993.

23D. OE, R. LP, J. HC, W. WRS, and J. CA, “AUTOMATIC DETECTION OF SEZURES USING ELECTROENCEPHALOGRAPHC SIGNALS,” US Patent, May 17, 1994.

24K. K. Jerger, T. I. Netoff, J. T. Francis, T. Sauer, L. Pecora, S. L. Weinstein, and S. J. Schiff, “Early seizure detection,” J Clin Neurophysiol, vol. 18, no. 3, pp. 259-68, May, 2001.

25L. D. Iasemidis, Epileptic Seizure Prediction and Control: IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, 2003.

26R. Meier, H. Dittrich, A. Schulze-Bonhage, and A. Aertsen, “Detecting epileptic seizures in long-term human EEG: a new approach to automatic online and real-time detection and classification of polymorphic seizure patterns,” J Clin Neurophysiol, vol. 25, no. 3, pp. 119-31, Jun, 2008.

27S. Belhadj, A. Attia†, B. A. Adnane, Z. Ahmed-Foitih, and A. T. Ahme, A Novel Epileptic Seizure Detection Using Fast Potential-based Hierarchical Agglomerative Clustering Based on EMD, IJCSNS International Journal of Computer Science and Network Security, 2016.