Site Loader

Electroencephalography
is a measurement device that observes the electrical activity of neuron by
using electrodes. “Epilepsy, commonly defined
as a neurological disorder, affects the brain, impacts about 2% of the world
population leading to a reduction in their productivity and also imposes restrictions
on their daily life”. By analyzing EEG signals
and patient behavior, one can diagnosis epilepsy easily. A lot of analytical
methods have been applied to EEGs in order to detect mental fatigue which is a
constant occupational hazard in daily life or simply fatigue (Fatemeh et al.,
2012). A seizure is the result of sudden excessive electrical discharges in a
group of brain cells. It occurs when the neurons generate uncoordinated
electrical discharges that spread throughout the brain. “Epilepsy is a recurrent seizure disorder caused by abnormal
electrical discharges from brain cells, describing the condition of the patient
having been recurring “spontaneous” seizures or brainstorm due to the sudden
development of synchronous firing in the cerebral cortex (Wu Ting et al., 2008; Tahir Ahmad et
al., 2012)”. The other clinical use of EEG is
in diagnosis of coma, brain death, and encephalopathy. Moreover, it can be used
in alcoholic consumption measurement, sleep stage detection, emotion
recognition, video quality assessment, measuring the change in the brainwaves
by smoking and mobile phone usages (Wu
Ting et al., 2008; Tahir Ahmad et al., 2012; Michal Prilepok et al., 2013;
Cecilia Damon et al.,2013; Shiliang Sunet al., 2014; Turkey et al., 2014).

A
kernel non-negative matrix (KNMF) based factorization for spectral EEG feature
extraction has been presented in (Hyekyoung Lee et al., 2009). A discriminative
spectral feature from time-frequency representation of EEG data was extracted
and experiments on two benchmarks EEG datasets confirmed the performance gain
over standard NMF. They also presented
multiplicative updates for KNMF which was kernelized version of the standard
version (Hyekyoung Lee et al., 2009). “A
comparison of GNMF with NMF and some modified NMFs, for spectral features from
EEG data was done. Experiments on brain computer interface (BCI) competition
data indicated that GNMF improved the EEG classification performance”. GNMF was useful in the task of subject-to subject
transfer where the prediction for an unseen subject was performed based on a
linear model learned from different subjects in the same group. A work based on
non-negative tensor factorization (NTF) for continuous EEG classification was
also proposed (Hyekyoung Lee et al., 2009). NTF approach was used to determine
discriminative spectral features using the Viterbi algorithm for continuous EEG
classification. NTF was useful in finding hidden structures for new dimensions
such as time and class Non-negative decomposition methods have been used to classify
fatigue from non-fatigue state using NMF, local non-negative matrix
factorization (LNMF), sparse non-negative matrix factorization (SNMF) and
discriminant non-negative matrix factorization (DNMF) (Hyekyoung Lee et al.,
2007; Fatemeh et al., 2012). “The EEG signals
were recorded from 32 channels and after two hours severe mental activity from
seventeen healthy subjects were also recorded”.
First, the preprocessing of raw EEGs was done. Later, they were arranged in
matrices for decomposition into discriminative subspaces. Support vector machine
(SVM) was used to classify the two mental states. Experimental results
demonstrated that discriminant NMF significantly (p<0.05) outperformed the other compared non-negative methods in terms of parameters namely, accuracy, feature storage, and robustness (Fatemeh et al., 2012). NTF technique for single channel EEG artifact rejection technique was proposed in 2003 (Cecilia et al., 2003). They informed source separation methods for artifact removal in EEG recordings with a low number of sensors, especially in the extreme case of single-channel recording, by exploiting prior knowledge from auxiliary lightweight sensors by capturing artifactual signals. To achieve this, they proposed a method using NTF in a Gaussian source separation framework that proved competitive against the classic Independent Component Analysis (ICA) technique. Both NTF and ICA methods were used in an original scheme that jointly processed the EEG and auxiliary signals. The adopted NTF strategy performed better in improving the source estimates accuracy in comparison to the usual multi-channel ICA approach (Cecilia et al., 2003). NMF along with a constraint has been implemented to increase the discriminability of two classes for feature extraction. Constraints used were placed on NMF and KNMF algorithms to increase the discriminability between two classes by increasing the energy difference between their potential sources in a spectral EEG signal. "The IDIAP database, which contained the motor-imagery related EEG spectrum of three subjects, was adopted to test the discrimination between two classes". Using this database, the classification accuracy of the proposed constraint was obtained 75%, which was 7% higher than what was obtained through NMF without a constraint. Also, the classification accuracy of KNMF with the proposed constraint was obtained 4% higher than that of KNMF without a constraint, and later reached 78% (Shiliang Sun and Zhou; Turkey; Motoki Sakai 2014). NMF has been used in decomposing inter trial phase coherence (ITPC) of multi-channel EEG (Morten Morup et al., 2006). Vamsi K. Potluru presented Group learning using contrast NMF: application to functional and structural MRI of schizophreniain was presented. NMF was used to learn the features of both structural and functional magnetic resonance imaging (sMRI/fMRI) data. NMF can be applied to perform group analysis of imaging data and to learn the spatial patterns which linearly covary among subjects for both sMRI and fMRI. An additional contrast term to NMF (called co-NMF) was added. Co-NMF was used for the identification of features distinctive between two groups. The approach was applied to a dataset consisting of schizophrenia patients and healthy controls. Classification rate of 55% and 68% were obtained and nearest neighbors algorithm was used for classification (Vamsi K et al., 2008).

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

Post Author: admin

x

Hi!
I'm Erica!

Would you like to get a custom essay? How about receiving a customized one?

Check it out