Time Frequency Image based features for detection of focal EEG signals

Abstract

In this paper, the Time-Frequency Image (TFI) of electroencephalogram (EEG) signals obtained from Time-Frequency Representation (TFR) is employed as a new method to analysis of focal EEG signals. Further, the gray image is sub-divided into frequency-bands known as rhythms of EEG signals. The image based features namely entropy, texture homogeneity (TH), inverse difference moment (IDM), power, and maximum count pixel intensity have been extracted from sub-images of EEG signals. Kruskal Wallis statistical test is performed for discrimination of focal and non-focal EEG signals. The lower probability value shows that the features have statistically significant for discrimination of focal and non-focal EEG signals.