A DCT based approach to epileptic seizure detection

Electroencephalogram (EEG) is a widely used tool for the clinical investigation of epileptic seizures. A new scheme of epileptic seizure detection using statistical features and Discrete Cosine Transform (DCT) is presented in this paper. Median absolute deviation (MAD) and variance is taken as the discriminating features between three different classes of EEG under study. The DCT was used for feature reduction, whose ability to pack input data into as few coefficients as possible makes it a good choice for the purpose. The representative DCT coefficients were given as the input to a linear classifier to yield 100% accuracy.