EEG based hearing threshold diagnosis using feed-forward network

In this paper, a simple analysis has been made to distinguish the normal and abnormal hearing subjects using acoustically stimulated EEG signals. Independent power spectral features of the brain rhythms (delta, theta, alpha, beta, and gamma) were extracted from the recorded EEG signals. The extracted power spectral features were then associated to the auditory perception and neural network models for the left and right ears were developed. The result indicates that the gamma-power derived from the electrodes can be deployed independently to characterize the EEG dynamics of normal hearing and abnormal hearing perception of a person. The effects of brain rhythms on perceiving two different auditory frequencies namely 500 Hz and 1000 Hz their associated auditory response have been investigated. From the network models, it has been inferred that the neural network models were able to discriminate the normal hearing and abnormal hearing persons.