Statistical Analysis for EEG Patterns Comparison Between Real Motion and Imagery Motion

Electroencephalogram (EEG) is a brain signal which widely used in the Brain-Computer Interface (BCI) system, where the data collection of EEG signals is done by a non-invasive method. Motor imagery recently has been an interesting issue since the development of many cognitive studies including neuromuscular disorder rehabilitation. This research is focusing on comparing two motions, one with real motion, and the other is with no motion, but by using imagery motion. C3 and C4 channels are used to record the EEG data. The data processing is done by filtering the EEG band by using Finite Impulse Response (FIR) to limit the low pass and high pass frequency. The raw data is then cleared from the artifact by using the Automatic Subspace Reconstruction (ASR) method. The clean EEG is then decomposed into sub-band alpha and beta using the Butterworth filter. For comparing the two kinds of motions, time-domain feature extraction is used with parameters: Mean, Mean Absolute Value (MAV), and Standard Deviation (STD). The analysis is carried out by comparing the results of extraction parameters of real motor motion with imagery motion. The experimental results show that STD produces the highest number of data frames with a similar value between the real motion and imagery motion compared to other parameters. The value of the STD extraction on the alpha band (mu-rhythm) frequency in C3 channel is 654 data frames (85.16%), and C4 channel is 672 data frames (87.45%), while the beta band frequency in C3 channel is 528 data frames (68.75%), and in C4 channel is 536 data frames (69.82%). The conclusion is standard deviation shows the best parameter with a similar pattern between real motion and imagery motion.