20 / 2016-06-28 17:34:31

A HYBRIDIZED NEURAL NETWORK AND OPTIMIZATION ALGORITHMS FOR PREDICTION & CLASSIFICATION OF NEUROLOGICAL DISORDERS

2603,2936,403

Artificial Neural Networks (ANN) have become the
most efficient method in biomedical areas for their effective
results in classifying several complex disorders. Application of
standard optimization techniques in combination with ANN
could highly optimize the parameters of this Network and make
it more reliable and efficient.
In this paper, a hybrid model of Artificial Neural
Network and Particle Swarm Optimization(PSO) Algorithm for
the Classification & Prediction of various Neurological Disorders
is designed. The proposed system works on the EEG signals
obtained from patients suffering from Focal Epilepsy, Brain
Death, Slow-wave conditions, etc. This single system is then
capable of performing Classification and Prediction of the
disease based on the EEG signal input. Here Probabilistic Neural
Network (PNN) is used as it is very efficient for classification
purposes. Prediction is performed by using Modified PSO. The
EEG database is obtained from CIIMS Hospital, Nagpur. The
results are highly reliable with graphs for Predicted signal and
Prediction Error and percentage of accuracy, sensitivity and
mean squared error are calculated as well. With the help of this
system classification of EEG signals into correct category of
disease can now be easily done with high accuracy and in a short
span of time.
This paper reviews a novel approach for classification and
prediction. Furthermore, this work will be helpful in future to
assist doctors in hospitals. As it is known that the EEG signals
are difficult to predict, it takes time for the doctors to analyze
them. But this work can prove to be time saving and can be
helpful for better diagnosis.