Journal of Ambient Intelligence and Humanized Computing | 2019
Developing brain abnormality recognize system using multi-objective pattern producing neural network
Abstract
According to the survey, brain abnormal mortality rate is increased up to 86% due to the severe effect of brain injuries, brain tumor, brain stork and other genetic mutations. The brain abnormality may occur in different disease such as Glioblastoma, Ependymoma/anaplastic ependymoma, bipolar disorder and so on. Due to the dangerous brain disease, it has to be detected in earlier stage for avoiding the mortality rate. So, in this paper introduce the multi-objective pattern producing neural network for developing the automatic brain abnormality recognizes system. Initially the brain signal such as Electroencephalogram (EEG) is collected from patient, noise present in the signal is removed using frequency normalization principal component analysis approach. The noise free signal is further examined, different features are extracted by applying ISO map spectral feature and particle bee based features are selected. The selected features are fed into the above-mentioned classifier that recognizes the brain abnormality related features according to the effective activation function. Then the efficiency of the brain abnormality recognize system is implemented in MATLAB tool and the excellence of the system is evaluated in terms of using error rate, sensitivity, specificity, F-measure, Mathew correlation coefficient and accuracy.