| dc.description.abstract | The fundamental movements like walking and speech communications are the basic needs of human beings in their daily life. Differentially Enabled (DE) communities have functional limitations that vary from primary/complicated activities such as weakening of the muscles, spasms, movement related problems, stroke, trouble swallowing and dysarthria. In these cases, Electroencephalogram (EEG) signals, being a measure of brain activity, which is responsible for the control of voluntary muscle movements through nervous system can be used to develop an alternative communication system or Brain Machine Interface’s (BMI). This research focuses on analyzing different frequency domain algorithms to recognize the navigational tasks along with communication tasks to develop an EEG based intelligent robot chair with communication aid (IRCC). IRCC involves classification of navigational tasks (Forward, Left, Right & Relax) and tasks related to speech (Yes, No and Help), to provide a navigation system with communication aid for the DE patients. Twenty healthy naive-, age-, and gender-matched normal subjects were participated in the data collection procedure while eight-channel wireless EEG signal was being recorded. Two simple data acquisition paradigms were implemented based on thought evoked potentials (TEP) and visually evoked potentials (VEP) to
determine the significant correlations between the brain dynamics and IRCC tasks. Furthermore, the developed database was analyzed in both customized modes and generalized modes. The recorded brain wave signals are preprocessed to remove the
interference waveforms and segmented into frame samples of equal length. The frame signals are categorized into six frequency band signals, (namely Delta, Theta, Alpha, Beta, Gamma-1 and Gamma-2), and used to extract the features. To classify the IRCC tasks, four distinctive frequency-domain feature extraction methods were compared (namely higher order spectra (HOS), cross correlation analysis, band power (BP) analysis and power spectral density (PSD)). Also, three different approaches based on crosscorrelation technique are proposed to estimate the interdependence between the frame signals, frequency bands and electrode positions (namely, statistical features using cross correlated two consecutive frames based spectral bands (CF), statistical features using cross correlated frame based combination of spectral bands (CFB), and statistical features
using cross correlated frame based combination of electrode channels (CEC). | en_US |
