International Journal of Engineering and Technology

This research article seeks to propose a Hilbert-Huang transform (HHT) based novel feature extraction approach for the analysis of multi-channel EEG signals using its local time scale features. The applicability of this recently developed HHT based new features has been investigated in the analysis of multi-channel EEG signals for classifying a small set of non-motor cognitive task. HHT is combination of multivariate empirical mode decomposition (MEMD) and Hilbert transform (HT). At the first stage, multi-channel EEG signals (6 channels per trial per task per subject) corresponding to a small set of non- motor mental task were decomposed by using MEMD algorithm. This gives rise to adaptive i.e. data driven decomposition of the data into twelve mono component oscillatory modes known as intrinsic mode functions (IMFs) and one residue function. These generated intrinsic mode functions (IMFs) are multivariate i.e. mode aligned and narrowband. From the generated IMFs, most sensitive IMF has been chosen by analysing their power spectrum. Since IMFs are amplitude and frequency modulated, the chosen IMF has been analysed through their instantaneous amplitude (IA) and instantaneous frequency (IF) i.e. local features extracted by applying Hilbert transform on them. Finally, the discriminatory power of these local features has been investigated through statistical significance test using paired t-test. The analysis results clearly support the potential of these local features for classifying different cognitive task in EEG based Brain –Computer Interface (BCI) system.