Error-based Analysis of VEP EEG Signal using LMS

Electroencephalography (EEG) involves the usage of electrodes placed on the human scalp to record electrical impulses generated by the brain. One of the many components that are present in EEG signals is the Visually Evoked Potential (VEP), whereby brief electrical impulses are generated as a result of the presence of visual stimuli. The aim of this project is to analyse EEG signals that contain VEP using the least-mean squares (LMS) method and differentiate between alcoholic and non-alcoholic subjects based on the resultant error signal. This LMS method is a form of adaptive filter that minimizes the mean square of the cost function for every iteration it undergoes and is widely used in many signal imaging applications due to its simplicity in implementation and low computational complexity. The EEG recording with VEP components is already available so the scope of the project only covers the adaptation of the LMS adaptive filter and the analysis of the VEP EEG error signals for 5 alcoholic and non-alcoholic subjects. The analysis of the results indicate that there is a certain range of standard deviation values in which it is possible to classify the condition of the subject into either alcoholic or non-alcoholic condition.
vii