1.

Automatic Removal of eye movement and blink artifact from EEG

data using blind component separation
Carrie A.Joyce, Irina F. Gorodnitsky and Marta Kutas

Problem
Signal from eye movement and blinks which included of artifacts from EEG data
can be order of magnitude larger than brain electrical signal. Removing the
artifacts by using blind component separation.
Methods
Using data recorded from 15 scalp and 6 EOG electrodes, referenced on-line to
the left mastoid and sampled at 500 Hz with an on-line band pass filter from 0.016
to 100 Hz. EOG data are preferable in ocular artifact detection. By using blind
source separation reduces mixtures of neural and non-neural variabel. BSS is
signal processing methodology includes independent component analysis.
Results
The signals originating from the eyes can be identified and removed from EEG
data. Blind component separation cleanly capture the artifact signals and separate
them from the components of brain activity.

2.
Removal of artifact from EEG signal using adaptive filter
through wavelet transform
P.Senthil Kumar, R.Arumuganathan, K.Sivakumar, C.Vimal

Problems
Removing electro-ocular (EOG) artifacts in the electroencephalogram (EEG)
records by using adaptive methods. Because the artifacts increase the difficulty in
analyzing the EEG signal.
Method
EEG recording are contaminated by EOG signal.

Applying wavelet transform to the contaminated EEG and reference EOG

Applying adaptive filter with LMS algorithm, the output signal is
subtracted from the corrupted EEG signal.
Applying wavelet reconstruction procedure to reconstruct the EEG signal
to produce the artifact free EEG signal.

Result

Figure 1. Adaptive filter

Figure 2. (a) Reference EOG
(b) corrupted EEG

The filter H(z) adapts the amplitude and phase of EEG + EOG artifacts. The
output signal y(n) is subtracted from the EEG contaminated with EOG artifact.
Adaptive filter based in LMS algorithm with wavelet decomposition were
described in order to cancel EOG artifacts. It can adapt their coefficients to
variation in heart frequency, abrupt changes in the line frequency or modification
due to eye movements. Adaptive cancellation with help of wavelet decomposition
is an efficient processing technique for improving the quality of EEG signal in
biomedical analysis.

3.
EEG Signal Classification Using Wavelet Feature extraction and Neural
Networks
Pari Jahankhani, Vassilis Kodogiannis and Kenneth Revett

Problem
Application of neural network models for classification of Electroencephalogram
(EEG) signals. The authors employed a discrete wavelet transform to the dataset
in order to extract temporal information in the form of changes in the frequency
domain over time. There are two stages of decision making : feature extraction
scheme using wavelet transform and learning-based algorithm classifier
performed classification.
Method
In this paper, the authors used the publicy available data described in Andrejak et
al. The complete data set consists of five sets (A-E). Sets A and B recorded on
five healthy volunteers. Volunteers were relaxed in an awake with eyes open (A)
and eyes closed (B). Set D recorded from within epileptogenic zone, set C from
hippocampal formation of the opposite of the brain. While sets C and D contained
only activity measured during seizure free intervals, set E only contained seizure
activity.
Result
The difference between normal and an epileptic EEG signal. The authors extracted
statistical information from the wavelet coefficients, which we used as input to a
set of supervised learning algorithms MLP and RBF based on neural networks.

Both classifiers were able to correctly map the inputs to the desired outputs with
appropriate and typical training periods-indicating that local minimal not a major
factor in training- although the RBF training period was substantially less than for
MLP.

4.
Removing Electroencephalographic Artifacts by Blind Source Separation
Tzyy-Ping Jung, Scoot Makeig, Colin Humphries, Te-Won Lee, Martin J.
Mckewn, Vicente Iragui, Terrence J. Sejnowski

Problem
Eye movements, eye blinks, cardiac signals, muscle noise and line noise present
serious problems for electroencephalographic (EEG) interpretation and analysis.
Proposed a new and generally applicable method for removing a wide variety of
artifacts from EEG record on blind source separation by independent component
analysis.
Method
The first EEG data set used in the analysis was collected from 20 scalp electrodes
placed and 2 EOG placements, all referred to the left mastoid. The sampling rate
was 256 HZ. The second data set was recorded at 29 scalp electrodes and 2 EOG
placements from an adult autistic subject. The third EEG data set contained 13
EEG channels ( no EOG channel) and was recorded at a sampling rate of 312.5
Hz.
ICA decomposition was performed on 10-s EEG epochs from each data set using
MATLAB 4.2c on a DEC 2100A/300 processor. The learning rate was gradually
reduced to 5 x 10-6 during 80 training iterations requiring 6.6 min of computer
time.
Result
ICA opens new and useful windows into many brain and non-brain phenomena
contained in multichannel EEG records by separating data recorded at multiple
scalp electrodes into a sum of temporally independent components. ICA is
generally applicable and effective method for removing artifacts form EEG
records because their time course are generally temporally independent and

spatially distinct from sources of cerebral activity. ICA has several advantages
compared with other artifact removal methods : (1) The algorithm is
computationally efficient and the computational requirements are not excessive
even for fairly large EEG data sets. (2) ICA is generally applicable for removal of
wide variety of EEG artifacts.
5.
Removal of Ocular Artifacts From Electro-encephalogram by Adaptive
Filtering

Problem
EEG signal especially those recorded from frontal channels, often contain strong
electro-oculogram (EOG) artifacts produced by eye movements. Existing
regression-based methods for removing EOG artifacts require various procedures
for preprocessing and calibration that are inconvenient and time consuming. The
paper describes a method for removing ocular artifacts based in adaptive filtering.

Method
EEG data were recorded from 12 sites positioned according to the international
10-20 electrodes. All signals were bandpass filtered at 0.48-30 Hz and the
sampled at 256 Hz. The noise cancelling using RLS algorithm.

Result
EEGs recorded at each of 12 channels, by using adaptive filter to remove EOG
artifact, RLS algorithm. The filter are basically shut down automatically (filter
coefficients become very small) and the original EEG simply passes the system
without visible changes.
There are two parameters that need to be determined : the forgetting factor and
the filter length M. The symbol, , can be considered as being associated with a
sample window that specifies how many previous samples are used to calculate
the current filter coefficients for producing the current output e (n).

Fig 1. EOG artifact removal using adaptive filtering (a) original (b) corrected (c)
horizontal EOG (d) vertical EOG

Fig 2. Comparison of results of adaptive filtering using three filter length : M=1,2
and 3
The perfomance of the adaptive filter is actually not very sensitive to the choice of
M.

The adaptive filter can automatically be adapted to a new condition without losing
effectiveness. By implementing a recursive algorithm, the filter is fast enough for
real-time processing.

6.
An Adaptive Method to Remove Ocular Artifact From EEG Signals Using
Wavelet Transform
P.Senthil Kumar, R.Arumuganathan, K.Sivakumar, C.Vimal

Problem
Removing ocular artifacts in the Electroencephalogram (EEG) records by using an
adaptive filter. The paper propose an adaptive filtering method, which makes use
of RLS (Recursive Least Square) algorithm for removing ocular artifactx form
EEG recordings through wavelet transform.
Method

Apply stationary wavelet transform to the contaminated EEG and

reference EOG
Apply adaptive filter with RLS algorithm
Apply wavelet reconstruction procedure to reconstruct the EEG signal to
produce the artifact free EEG signal

The data is sampled at a rate of 128 samples per second.

Result
Adaptive filter with RLS algorithm through wavelet transform reduces the
artifacts in EEG and is an efficient processing technique for improving quality of
EEG signal.

7.
Comparison of Independent Component Analysis Algorithm for Removal of
Ocular Artifacts from Electroencephalogram
V. Krishnaveni, S. Jayaraman, P.M.Manoj Kumar, K.Shivakumar, K.Ramadoss

Problem
EEG recordings are distorted by electroouculogram (EOG) artifacts causing
serious problem for EEG interpretation and analysis. An preferable method is to
apply Independent Component (ICA) algorithms like OGWE, MS-ICA, SHIBBS,
Kernel-ICA, JADE and RADICAL are compared and to prove where the best ICA
algorithm at separate the source signal from mixed EEG is.
Method
ICA algorithm like OGWE, MS-ICA, SHIBBS, Kernel-ICA, JADE, and
RADICAL are assessed and compared by mutual information(MI) estimator based
on k-neighbor statistics without using the probability density functions. For
removal of ocular artifacts from EEG (32 electrodes, 128 Hz).
Result
RADICAL algorithm performs best at separating the source signal from EEG
signal (mixed with ocular artifacts).

Fig 1. Mixed EEG signal

Fig. 2 Independent Components

obtained form RADICAL

Fig 3. Contaminated and corrected EEG for different ICA algorithm

8.
Removal of Ocular Artifacts in the EEG through Wavelet Transform wihtout
using an EEG Reference Channel
P.Senthil Kumar, R.Arumuganathan, K.Sivakumar, C.Vimal

Method
This paper proposed a statistical method for removing ocular artifacts in the
electroencephalogram (EEG) records through wavelet transform without using
and EOG reference. Noise source increase the difficulty ini analyzing in the EEG
and obtaining clinical information. It necessary to design a procedure to decrease
such artifacts in EEG records.
Method
EEG data with ocular artifacts are taken from
http://www.sccn.ucsd.edu/~arno/famzdata/publicly_available_EEG_data.html .
The data is sampled at a rate of 128 samples per second. The effect of ocular
artifacts will be dominant in the frontal and frontopolar channel like FP1, FP2, F7,
and F8.

Apply stationary wavelet transform to the contaminated EEG signals

Identify the spikes in the contaminated EEG
Identify the ocular artifact spike zones using coefficient of variation, a
statiscal approach
Apply de-noising technique- to fix the suitable treshold value and
threshold function for the artifact zones
Apply wavelet reconstruction procedure to reconstruct the EEG signal

Result
The proposed method is to minimize the complexity of the work and easily
identify the artifact zones for removing the artifacts. The proposed method gives
less complexity and easier to removal of the artifacts with the help of wavelet
decomposition and is an efficient technique for improving the quality of EEG
signals in biomedical analysis.

Fig 1. (1) EEG with artifact (2) Modified threshold function (3) threshold function
(4) proposed method

9.
Artifact Removal from EEG signals using adaptive filters cascade
A Garcs Correa, E Laciar, H D Patio, M E Valentinuzzi

Problem
Artifact in EEG records are caused by various factors, like line interference, EOG
(electrooculogram) and ECG (electrocardiogram). These noise sources increase
the difficulty in analyzing the EEG and to obtaining clinical information. The
paper proposed a cascade of three adaptive filters based on a least mean square
(LMS) algorithm.
Method
Five EEG records were filtered with the proposed adaptive cascade filter. The
input signal is the corrupted EEG . Three reference are line-frequency, ECG, and
EOG.

Fig 1. Adaptive filters cascade. The input signal corrupted with line-frequency,
ECG and EOG.

Result
Three adaptive filters in cascade, based on LMS algorithm, were described in
order to cancel common artifacts ( line interference, ECG and EOG ). The
advantages of using a cascade of three filters instead of filtering the three signals
with a single adaptive filter are among others :

The coefficient`s adaptation in three independent filters is simpler and

faster than their adaptation in a single filter
At each stage output, EEG with one of the three attenuated artifacts are
present; such separation (by artifact) may be useful in some applications
where such output might be enough

In all cases, artifact were adequately attenuated, without removing significant

useful information. We conclude that adaptive cancellation is an efficient
processing technique for improving the quality of EEG signals in biomedical
analysis.

10.
Removal of Ocular Artifacts from EEG Using an Efficient
Neural Network Based Adaptive Filtering Technique
S. Selvan and R. Srinivasan

Problem
Ocular artifacts act as major source of noise, making it difficult to distinguish
normal brain activities from the abnormal ones. An efficient technique that
combines two popular adaptive filtering technique, namely adaptive noise
cancellation and adaptive signal enhancement, in a single recurrent neural network
is proposed for the adaptive removal of ocular artifacts from EEG.

Method
The schematic diagram of the proposed technique is on the figure below.

The network consist if six nodes where node 1, node 2, and node 3 are

computational node and node 4, node 5 and node 6 are input nodes. The noise 1
and noise 2 come from EOG signal for vertical and for horizontal eye movements
are fed to node 6 and node 5.