DEPRESSION LEVEL PREDICTION USING EEG SIGNAL PROCESSING

Depression is one of the most common mental disorders that at its worst can lead to suicide. Diagnosing depression in the early curable stage is very important. It may also lead to various disorders like sleep disorders and alcoholism. Here in this project the Electroencephalogram Gram (EEG) signals are obtained from publicly available database are processed in MATLAB. This can be useful in classifying subjects with the disorders using classifier tools present in it. For this aim, the features are extracted from frequency bands (alpha, delta and theta). Primarily the EEG signals were read using EDF browser software and the signals were loaded into Matlab to get log Power Spectral Density from EEG bands. The results obtained from Matlab are fed into neural network pattern recognition tool and ANFIS tool box which is integrated in MATLAB. These are powerful tool for data classification. Relevant extracted features parameters are used as inputs to the ANFIS and nprtool. The evaluated outputs are helpful to distinguish alcoholics from controls and various sleep disorders like insomnia, narcolepsy, bruxism and nocturnal frontal lobe epilepsy. 20 samples are trained and evaluated for Alcoholism and 40 samples are trained and evaluated for 4 different sleep disorders in ANFIS tool. The evaluated ANFIS output is read as 0 for Insomnia, 1 is for No sleep disorder, 2 for Narcolepsy, 3 for NFLE, 4 for Bruxism. 240 samples for 4 different sleep disorders and 60 samples for Alcoholism/ Control are trained and classified in nprtool.