Contents Acknowledgment Abstract List of Figures List of Tables List of Abbreviations 1

List of Figures
List of Tables
List of Abbreviations
1. Introduction
1. Aim of this thesis
2. Thesis framework
2. Background, History & Components of EEG
1. History
2. Elementary concept of electroencephalography
3. How EEG is examined?
4. Classification of EEG signal or EEG waveform
5. What do the readings show?
6. Interpretation of EEG results
7. Components of EEG machine
7.1 EEG cap
7.2 Electrodes
7.2.1 Electrode placement
7.2.2 Montages
7.4Computer control module and display device

3. Methodology
1. Fourier transform
2. Wavelet transform
3. Classification of wavelet transform
Discrete wavelet transform
Continuous wavelet transform
4. Some common wavelets
The Haar wavelet
The Mexican hat wavelet
The Morlet wavelet
5. Wavelet power spectrum
6. Global wavelet spectrum
4. Results and Discussions
5. Applications of EEG
6. References
List of Figures
2.1Image courtesy of
2.2Delta wave
2.3Theta wave
2.4Alpha wave
2.5Beta wave
2.6Gamma wave
2.7Needle electrodes
2.8Surface electrodes
4.1(a) EEG signal; (b) The wavelet power spectrum using Morlet mother wavelet;
(c) The global wavelet power spectrum; and (d) Scale-average wavelet power.
Wavelet power decreases according to the colour order as: red, orange, yellow, blue and

4.2(a) EEG signal; (b) The wavelet power spectrum using Morlet mother wavelet;
(c) The global wavelet power spectrum; and (d) Scale-average wavelet power.
Wavelet power decreases according to the colour order as: red, orange, yellow, blue and

4.3(a) EEG signal; (b) The wavelet power spectrum using Morlet mother wavelet;
(c) The global wavelet power spectrum; and (d) Scale-average wavelet power.
Wavelet power decreases according to the colour order as: red, orange, yellow, blue and

Basic Definitions
Shrinking or stretching of original figure
The term translation is used to move an object a fixed distance without altering size and shape of original figure.

Chapter 1
Brain is the most important organ of human body. It is interconnected with billions of nerve cells, called neurons which make millions of new connections and communicate with each other by passing electrical impulses due to which Signals are formed within individual neurons. The pattern and strength of these connections constantly change 1.
There are many methods to measure electrical or metabolic activity of the brain, out of which electroencephalography is one of the most widely used method because of its high temporal resolution 17. The other methods for recording brain signals are Functional Magnetic Resonance Imaging (FMRI), Near Infra-Red Spectroscopy (NIRS), Magnetoencephalography (MEG), Electrocorticography (ECoG) and Positron Emission Tomography (PET) 17 “1.3 components of a BCI system”
EEG is a source to investigate brain diseases such as epilepsy, autism spectrum disorder (ASD) or to report the presence of brain tumors. There is a significantly different patterns in EEG signal of epilepsy patients during a seizure compared to the normal state of the brain with respect to space, time and frequency patterns “ch 2 epilepsy and eeg”.

EEG signals are non-stationary in nature. For the analysis and representation of non-stationary EEG signals, three most common methods are used to get the desired information. The common methods are Fourier transform, Short-Time Fourier transform and Wavelet transform. More details about these methods are described in chapter 3. This representation can be used to characterize transient events, reduce noise, compress data, and perform many other operations 26
1. Aim of this thesis
The purpose of this thesis is to propose a method of automatic detection of epileptic seizure using wavelet based features, in which normal and epileptic EEG signals are distinguished using wavelet power spectrum, global spectrum and average variances.

2. Thesis framework
This thesis is designed as follows:
Chapter 2 has a brief description of electroencephalography. In this chapter history, equipment of EEG and further details are covered.

Chapter 3 introduces the applied methodology.

Chapter 4 results and discussions are followed.

Chapter 5 A summary of the thesis together with a conclusion and suggestions for further work is presented in this chapter.

Chapter 2
Background, History and Components of Electroencephalography (EEG)
1. History
For the first time electrical impulses was observed in July 1875 by Richard Caton (1842-1926) through the galvanometer, which is a type of electrical ammeter to measure electric current. Later on Ernst Von Fleischl- Marxow put forward his studies on evoked potentials, linking nervous system activity to muscle movements.
In 1920, German physiologist and psychiatrist, Hans Berger (1873-1941) used these two discoveries as the foundation of what he would soon discover, the electroencephalogram (EEG). He first recorded Alpha (bottom picture) and Beta (top picture) waves 2. He used his ordinary radio equipment to amplify the electrical activity of human’s brain. He observed that the changes occurred in the electrical activity of brain showed different functional status of the brain, such as in sleep, anesthesia and in certain neural diseases, such as in epilepsy etc. 3.

Fig: 2.1 Image courtesy of 2
In 1934, epileptiform spikes was first demonstrated by Fisher and Lowenback. Interictal spike waves and the 3 cycles/s pattern of clinical absence seizures are specified by Gibbs, Davis and Lennox in 1935. Later on, in 1936 Gibbs and Jasper reported that the interictal spikes are the principle mark of epilepsy.

In 1947, The American EEG Society was founded and the first International EEG congress was held.

2. Elementary Concept of Electroencephalography (EEG)
Electroencephalography (EEG) is a non-invasive technique used to record electrical activity (normal or abnormal) generated by the brain. It records brainwaves pattern which helps to diagnose and evaluate brain disorders such as head injuries, brain tumors and various disorders like epilepsy, autism etc. EEG activity is quite small, measured in microvolts.

It is a valuable tool to measure and graphic display of a voltage fluctuation recorded over time resulting from ionic current within the neurons of the brain 4. For this purpose, EEG caps are used to place on a subject’s head which consist of electrodes to measure the electrical impulses or directly electrodes can be placed on a subject’s head by using gel which serves as a conductive media. Electrodes are crucial to measuring the electrical charges.

3. How EEG is examined?
The following processing steps are taken out for examining the EEG waves:
EEG recording
Experimental paradigm
Data triggering and data class assignment
Artifact control (Eye-induced artifacts, ECG (cardiac) artifacts, EMG (muscle activation)-induced artifacts, Glossokinetic artifact)
Feature extraction by means of spatial filtering (common spatial patterns)
Data classification
4. Classification of EEG Signal or EEG Waveforms
The EEG signal indicates the electrical activity of the brain. The electrical signals from the brain are picked up by small electrodes (about one centimetre across), which are placed on the person’s head. These waves are generally categorized on the basis of signal frequencies. The main frequencies of the human EEG waves are follows:
Delta (below 3.5 Hz): This activity is found in deepest meditation and dreamless sleep 5. They tend to have a relatively high amplitude in comparison of other frequency ranges 15. If there is abnormal delta activity, an individual may suffer with learning disabilities or have difficulties in maintaining conscious awareness 6.

Fig: 2.2 Delta wave
Theta (4 to 7.5 Hz): It is normally seen in young children in the frontal or frontalcentral head regions from 6 to 7 Hz. The appearance of frontal theta can be facilitated by emotions, focused concentration and during mental tasks 4.

Fig: 2.3 Theta wave
Alpha (8 to 12 Hz): It is the first rhythmic EEG activity seen by Hans Berger. That’s why these waves are also named as Berger’s wave. This activity is noted in the both side’s posterior regions of the head, being higher in amplitude on the dominant side 7. Alpha wave is the frequency range between Beta and Theta waves 6. These brain waves are existed in deep relaxation and usually seen in closed eye situation 8.

Fig 2.4: Alpha wave
Beta (above 13 Hz): This activity is closely associated with motor functions. It may be shortened or absent in areas of cortical damage 7. Too much beta activity may lead to experience excessive stress and/or anxiety 6.

Fig 2.5: Beta wave
Gamma (): These waves are associated with learning, thoughts and information processing 6.

Fig 2.6: Gamma wave
5. What do the readings show?
The EEG records the electrical activity of the brain. The type of brain activity that happens depends on many different things: if the person is awake or in different stages of sleep, what they are doing, and if their eyes are open or closed. Some activity seen in ‘well’ children would not be expected to be seen in healthy adults

6. Interpretation of EEG Results
The EEG test results are interpreted by feature extraction based on discrete wavelet transform (DWT). DWT has been adopted to extract various features. Once you have your EEG data, utilize acknowledge software to run a number of fully automated EEG analysis function. A range of fully automated routines provide quick easy and reproduce able results and include approximate entropy, delta power analysis, derive alpha RMS, EEG frequency band and frequency analysis. The system uses power spectral density to report mean power, median frequency, mean frequency 16.

7. Components of EEG machine
In 1924 Berger was the first to record human brain activity through EEG. Berger’s first recording device was very primitive. He fitted silver wires under the scalp of his patients. Later on, these were replaced by silver foils which was attached to the patient’s head by rubber bandages. Berger connected these sensors to a Lippmann capillary electrometer, with unexpected results. However, more accurate measuring devices, such as the Siemens double-coil recording galvanometer, which exhibit electric voltages as small as one ten thousandth of a volt, led to success.

The basic systems of an EEG machine contain data collection, storage, and display. The components of these systems are:
EEG cap
Electrodes with connecting wires
Computer control module and display device.

7.1 EEG Cap 
EEG caps of suitable fabric can be manufactured in this way that it is easy to fit and can be sized accurately. Electrodes in EEG caps can be placed through embedding, knitting or weaving. The fabric cap has recessed tin electrodes attached to the lycra-type fabric or composite fabric.

7.2 Electrodes
An electrode is an electrical conductor used to provide current through nonmetal object to measure conductivity. There are many types of electrodes for acquiring data for interpretation. Following are the types of electrodes with different characteristics used in an EEG machine:
Wet electrodes
Electrolytic gel is used to paste wet electrodes to make their contact with skin, penetrate hair and provide a clean conductive path. The Ag/AgCl electrodes are most commonly used as a wet electrodes.

Dry electrodes
There is no need for conductive gel for dry electrodes. It has potential for sensitivity reduction to motion artifacts and enhancement of signal to noise ratio.

Needle electrodes
In general, needle electrodes provide greater signal clarity because they are injected directly into the body. This eliminates signal muffling caused by the skin.

Fig 2.7: Needle electrodes
Surface electrodes:
For surface electrodes, there are disposable models such as the tab, ring, and bar electrodes. There are also reusable disc and finger electrodes. The electrodes may also be combined into an electrode cap that is placed directly on the head 9.

Fig 2.8: Surface electrodes
Textile electrodes:
The type of electrodes which is made from fabric is textile electrode. These electrodes do not required gel to make contact to the skin due to their conductive property. The textile electrodes can be manufactured by knitting, weaving or embroidering conductive yarn to the fabric.

The textile electrodes are good for a long time measurement because they do not cause irritation on the skin.

7.2.1 Electrode Placement
Electrodes are placed according to international 10-20 or 10-5 system. It is an international system to specify the electrode location in scalp which have been provided by the American Encephalographic Society (1994) as well as Oostenveld ; Praamstra (2001). In the 10-20 system, electrodes are placed at 10% and 20% points along lines of longitude and latitude.
Each site has a letter to identify the lobe and a number to identify the hemisphere location. The position of electrode placement is important in order to record data from the desired location. Different types of activities can be identified by these location over certain areas of brain. Letters and numbers with their position and functions are as follows:
Letters ; numbersElectrode’s Site with Function or Activities
F (frontal)F7 near centers for rational activities
Fz near intentional and motivational centerss
F8 close to sources of emotional impulses
C (central)C3, C4 and Cz deal with sensory and motor function
T (temporal)T3 and T4 for emotional process
T5 and T6 certain memory functions
O (occipital)O1 and O2 for primary visual areas
P (parietal)P3, P4 and Pz contribute to activity of perception & differentiation
Even numbersrefer to electrode position on the right hemisphere
Odd numbersrefer to those on the left hemisphere
z (zero)refers to an electrode placed on the midline

A: Left View of HeadB: Top view of Head

C: Location and nomenclature of the intermediate 10% electrodes, as standardized by the American Electroencephalographic Society. (Redrawn from Sharbrough, 1991.).

Fig 2.9:
7.2.2 Montages
The pattern of logical representation and orderly arrangement of electrode derivatives for multiple channels recorded simultaneously to enhance recognition of EEG waveforms.

The common montages need to be used for recording EEG waveforms are as follows:
Bipolar Montage
The term Bipolar refers that two electrodes per one channel. Each channel represents the voltage difference between two nearby (adjacent) electrodes, either anterior to posterior (longitudinal bipolar) or side to side (transverse bipolar) 6 14.

Referential Montage
In this EEG montage, each channel represents the difference between a certain electrode and a selected reference electrode. In referential montage there is a common reference electrode for all channels 10. Commonly, the Cz electrode is chosen as a reference electrode 14.

Average Reference Montage
The outputs of all of the amplifiers are summed and averaged. Frontopolar, frontal and occipital electrodes are often omitted from this montage to reduce contamination 11.

Laplacian Montage
The difference between an electrode and a weighted average of the nearest surrounded electrodes is refer to as Laplacian montage or source derivation montage 14.

7.3 Amplifier
As the signals travel through the machine, they run through amplifiers that make them enlarge to be clearly manifested. In EEG machine the amplifiers are designed to serve the two main functions
Differential discrimination
Each electrode is connected to one input of a differential amplifier (one amplifier per pair of electrodes); a common system reference electrode is connected to the other input of each differential amplifier. These amplifiers amplify the voltage between the active electrode and the reference (typically 1,000–100,000 times, or 60–100 dB of voltage gain).

Computer control module & Display device
Normally medical grade computers are preferred to be used which are tested and confirmed to be used in medical equipments. Specialized medical software is used to analyze, display and provide special functions and options which can be helpful in diagnosis of epilepsy. › Volume.
The EEG machine then either prints out the wave activity on paper (by a galvanometer) or stores it on a computer hard drive for display on a monitor.
Chapter 3
The individual’s brain areas possess unique information at any given time. An EEG signal consists of many waves with different characteristics as described in chapter 2. It is difficult to interpret the large amount of data obtained from even a single EEG electrode pair. Signal processing methods are needed to automate signal analysis and interpret the signal phenomena 18.

There are many signal processing techniques which have been widely used to extract the features from EEG signals or to transform the signal into other representation 19 21. Some of common methods are Fourier transform, Short-Time Fourier transform (STFT), eigenvector methods (EM), wavelet transform (WT), and auto regressive method (ARM), and so on 19. Each of these methods has its own property. Some methods only give temporal information, some methods are used for getting only frequential information while some particular methods provide time-frequency representation.
Some of these methods are described in this chapter. Out of which we focused only wavelet transform in details as per this thesis requirement.

1. Fourier Transform
Fourier series is the commonly known and historically first method used for decomposition of a signal into a (possibly infinite) sum of sin and cos functions with specific frequention and phase 20. The Fourier transform represents a signal from time domain to frequency domain without giving any information that at what time frequency is occurred 21.

A signal can be either continuous or discrete, and it can be either periodic or aperiodic. These two features generates four categories which are as follows:
These signals extend to both positive and negative infinity without repetition of periodic pattern. The Fourier transform of this type of signal is simply called the Fourier Transform 22.

The version of the Fourier transform in which waveform repeats itself in a regular pattern from negative to positive infinity is called Fourier Series 22.
This is the type of signals which are defined at discrete points between positive and negative infinity, and do not repeat themselves in a periodic manner. In this case Fourier transform is called Discrete Time Fourier Transform 22
These are discrete signals having a periodic fashion from negative to positive infinity. In this case Fourier transform is sometimes called the Discrete Fourier Series, but is most often called the Discrete Fourier Transform 22.

2. Wavelet Transform
The word wavelet first introduced by a geophysical engineer Jean Morlet 23. Wavelets are mathematical functions used to split up data into different frequency components, and then study each component with a resolution matched to its scale 24. Wavelet transform, similar to Fourier transform decomposes the signal onto a basis of function 25.
Wavelet transformation simultaneously provides the information about frequency of the signal and time at which frequency occurred by decomposing or transforming a one-dimensional time series into a diffuse two-dimensional time-frequency image
It is becoming a powerful signal processing tool for analyzing localized variations of power within a time series. However, the wavelet transform can be used to analyze time series that contain non-stationary power at many different frequencies (Torrence ; Compo, 1998).

The wavelets are generated from a basic wavelet function ?t ? L2(R) with zero mean in both time and frequency is defined by 32 21 33 34:
?j,kt=2j2 ?(2j2t-k)(3.1)
Where j, k are arbitrary integers which dilate and translate the function and ?j, k is an orthonormal basis in a Hilbert Space L2R 33 34.

2.1 Conditions of wavelets
(1) Average value of the wavelet must be zero:
-?+??(t)dt=0(2) The integral of the square of ?(t) is unity:
-?+??(t)dt=1(3) Admissibility condition:
If ?t has the Fourier transform ?? which is defined as:
??=-?+??te-i(2??)t dtThen the following must hold:
C?=0+???2?d? ; +?Here ?? is the Fourier transform of ?t and C? is admissibility constant.

2.1 Types of wavelet transform
There are two main types of wavelet transform:
i) Discrete wavelet transform: It deals with the discretely sampled function or time series 28. It is commonly used for pattern recognition and image compression and use specific subset of all scale and translation values 27.

ii) Continuous wavelet transform: It works for a signal that defines over a continuous time and operates the signal over every possible scale and translation values 27 28. Furthermore, CWT can be defined as the sum over all time of the signal multiplied by scaled, shifted version of the wavelet function ?. Mathematically 29,
Where x(t) represents a signal or time series, asterisk denotes the conjugate complex value, ? and s represents the time dimension and the scale dimension, respectively, and ?(t) represents the Morlet wavelet function 31.

2.2 Some common wavelets
There are many different wavelets with different properties like some of them are good for signals with sharp edges and some are for smooth signals 21. Some common wavelets are defined below:
The Haar wavelet
This wavelet is one of the oldest and simplest orthonormal wavelet function, named after A.Haar 21 28. It is a kind of continuous wavelet transform with discrete coefficients 30. Mathematically it can be expressed as a step function ?t 21:
?t=1 0?t;12,-1 12?t;1,0 otherwise.(3.3)
The Mexican hat wavelet
The Maxican hat wavelet is the Gauss’s second derivative. All derivatives of the Gaussian distribution function can be used as wavelet 21. Mathematically,
?t=(1-t2)e-t22 (3.4)
Where e-t22 is the Gaussian distribution function.

The Morlet wavelet
The Morlet wavelet is the most commonly used complex wavelet. It is defined as:
f0 is central frequency and ?-14 ensures that the wavelet has unit energy 21.

2.3 Wavelet Power Spectrum
Wavelet power spectrum allows to determine the distribution of energy within the data array. Where large power in WPS allows to determine that which features of signals are important to compute and which are not.

Mathematically it is defined as the absolute squared value of wavelet transform coefficients
Ws22.4 Global wavelet spectrum
Chapter 4
Results and Discussions
In this chapter, results of time series analysis of EEG signals of epileptic patient and healthy persons are compared by applying wavelet approach mentioned in chapter 3. The properties of epileptic and non-epileptic are significantly different from each other.
4.1 Data
The dataset used for this study has been collected from——-. The whole data consists of five EEG datasets from set A to E. Each dataset has 100 single channel EEG signals with duration of 23.6 seconds with a sampling rate of 173.61 Hz. The dataset is classified into five different classes i.e. Z, O, N, F and S. Three sets are selected for this study which are A, B and C. Set A represents Z001, set B represents O001 and set C represents S001 where Class Z shows—– state, class O is ——- and class S is ——– state.

4.2 Wavelet power spectrum
The parameters for the wavelet analysis are set as dt=1, s0=2*dt, dj=0.25, j1=10dj in order to do 10 powers-of-two with dj sub-octaves each. Figures 1(b)-3(b) is showing the power i.e. absolute value squared of the wavelet transform for the EEG data presented in figures 1(a)-3(a). The curve in figure 1(b)-3(b) indicates the 95% confidence interval. Values below the curve are not statistically significant and will not be considered in the analysis.

4.3 Global wavelet power spectrum
Figures 1(c)-5(c) shows the frequency of time series by an integration of power over time which shows the strongest peak at particular period. In figure 1(c), there are three prominent peaks but the most important peak is around period 128. The peak around period 1024 is below the confidence level that’s why it will not be considered in analysis. In figure 3(c), the highest peak is at period 32 and two other peaks are around period 8 and 64.

b) Wavelet power spectrumc) Global wavelet spectrum

d) Scale average time series

Figure 4.1. (a) EEG signal; (b) The wavelet power spectrum using Morlet mother wavelet; (c) The global wavelet power spectrum; and (d) Scale-average wavelet power. Wavelet power decreases according to the colour order as: red, orange, yellow, blue and white.


Figure 4.2. (a) EEG signal; (b) The wavelet power spectrum using Morlet mother wavelet; (c) The global wavelet power spectrum; and (d) Scale-average wavelet power. Wavelet power decreases according to the colour order as: red, orange, yellow, blue and white.


Figure 4.3. (a) EEG signal; (b) The wavelet power spectrum using Morlet mother wavelet; (c) The global wavelet power spectrum; and (d) Scale-average wavelet power. Wavelet power decreases according to the colour order as: red, orange, yellow, blue and white.

Chapter 5
Applications of EEG
The latest advancements in computer hardware and processor technology have empowered researchers to vastly expand the existing knowledge about the complexity of the human brain and gain deeper insights into brain processes and structures.

EEG can be used for various applications. Below is a list of the six most common applications of EEG technology:
1.1 Neuromarketing:
Neuromeasurement tools, such as EEG and GSR are a natural complement to eye tracking and present the next step of innovation in behavioural research 12.

In the field of neuromarketing, economists use EEG research to observe brain processes that pickup consumer decisions, brain areas that are active during product purchasing or providing services, and mental states of respective person during exploring physical or virtual stores.

A predictive modeling framework is proposed to understand consumer choice towards E-commerce products in terms of “likes” and “dislikes” by analyzing EEG signals 13.

1.2 Neuroergonomics(Human factors):
Neuroergonomic studies rely on neuroimaging techniques to understand brain structures, mechanisms, and functions during work.

Originating from Psychology, the field of Human Factors focuses on workplace optimization; both with respect to tools and interfaces as well as social interaction. In this area, EEG research is used to identify brain processes related to specific personality traits such as intro-/extroversion or social anxiety.

Additionally, brain processes reflecting cognitive and attentional states during human-machine-interaction are heavily studied using EEG, primarily using wireless headsets with long-term monitoring capabilities.

1.3 Social cooperation:
Humans are social agents spending most of time interacting with others. In social interaction research, brain processes are examined related to social image, self-assessment, and social behavior.

To study the brain processes underlying the occurence of communication and actions, EEG researchers use a method referred to as “hyperscanning” to record data from various people at once, allowing them to gain deeper intuition into leadership and team interactions
1.4 Psychology and Neuroscience:
Most generally, psychological studies utilize EEG to study the brain performance underlying attention, learning, and memory.

Based on massive time locked experimental trials, EEG researchers extract event-related potentials (ERPs) from EEG data, which allows characterizing brain processes caused by the events on a very detailed timescale (tens of milliseconds).

ERPs can be characterized by their amplitude (in millivolts, with positive and negative going waves labeled “P” and “N”, respectively), timing (in ms relative to event onset), and voltage distribution across all electrodes (topography).

1.5 Clinical and Psychiatric Studies:
Whenever brain processes are impaired (e.g., lesions, genetic dysfunctions, diseases), behavioral ; cognition disorders can be observed. Clinical and psychiatric fields use EEG to diagnose and monitor seizure disorders.

1.6 Brain Computer Interfaces (BCI):
A relatively new but emergent field for EEG is brain-computer interfaces. Today, we know in much more detail about brain areas which are active during perceiveing stimuli, executing bodily movements, or learning and memorizing things.

This gives rise to very powerful and targeted EEG applications to drive devices using brain activity. This can, for instance, help paralyzed patients steer their wheelchairs or move a cursor on a screen, but BCI technology is also used for military scenarios where soldiers are equipped with an exoskeleton and EEG cap, allowing them to move, lift and carry very heavy items simply based on brain activity. Further details are discussed in next chapter.
Chapter 6
1 introduction: the human brain by Helen Philips
2 history of eeg
EEG signals are a reflection of individual-dependent inner mental tasks.

5 https://brainworksneurotherapy.com6 5 types of brain waves frequencies: Gamma, Beta, Alpha, Theta, Delta;Brain7 Electroencephalograhy
8 this is how brain waves contribute to the state of mind;brain –wave

9 EEG:;eeg_n
11 how to record eeg ›
13 Analysis of EEG signals and its application to neuromarketing AuthorsMahendra Yadava ,Pradeep Kumar,Rajkumar Saini ,Partha Pratim Roy ,Debi Prosad Dogra
14 An introductory text and Atlas of Normal ; Abnormal findings in Adults, Children and infants: Erik K. St. Louis, MD
15 Develop an Automated system for EEG Artifacts Identificatiion.

16. fundamental of eeg recording july 6, 2017, life science data
17″ feature extraction of eeg signal using wavelet transform”.
18 EEG Signal Processing: Theory and Applications Nitish V. Thakor David L. Sherman
19 Methods of EEG signal features using linear analysis in frequency and time-frequency domains
20 Introduction to brain computer interface bachelor’s thesis 2012 (Dominik Zajicek)
21 wavelet transforms and efficient implementation on the GPU. Master thesis (Hanne Moen). May 2, 2007
22 The scientist and engineer’s guide to Digital Signal Processing by Steven W. Smith, Ph. D. ch 8: the discrete fourier transform
23 wavelet-based methods for the analysis of fMRI time series. Wink, Alle Meije (2004).

24 an introduction to wavelets by Amara Graps
25 Analysis of real time EEG signals by Sangeetha Munian, Sivakumaran Sivalingam and Vinoth Jayaraman
26 Small Wind Turbine Generator Condition Monitoring: Test Rig and Preliminary Analysis by Hongwei cai
28 Wavelet Neural Networksand their application in the study of dynamical systems by David Veitch, august 2005
29 wavelet toolbox for use with matlab by Michel Misiti, Yves Misiti, Georges Oppenheim and Jean-Micheal Poggi
30 Term paper A Tutorial of the Morlet Wavelet Transform
31 Continuous Wavelet and Hilbert-Huang Transforms Applied for Analysis of Active and Reactive power consumption available from: - accessed Aug 29 2018.

32 Wavelet transform, Sheng Y
33 A mathematical introduction to wavelets, P.Wojtaszczyk
34 An introduction to wavelets for economics by Christoph Schleicher