Motor imagery-based Brain-computer interface (BCI) is a control system that enables paralyzed people to communicate with the surrounding via their brain signals. In order to design such system, a powerful algorithm is required that translate the patient’s thoughts into control signals for commanding external gadgets. In the past, various powerful algorithms have been proposed to improve the efficiency of the BCI systems. These algorithms have two major drawbacks: firstly, most algorithms rely on features extracted from multi-channel data acquisition systems. The more the number of channels more is the complexity of algorithms and computational time, which reduces the system’s efficiency. Secondly, several algorithms deal with the extractions of brain activities and their classifications. They did not focus on the dynamical behavior of the brain activities.
At present, the research is focused primarily on optimizing these two drawbacks of the existing systems. The main objective of current research is to use the minimum number of channels to extract maximum information from the brain. To progress in these directions, both linear and graphical methods were employed to explore the brain activities during motor execution (ME) and motor imagery (MI) movements. The linear methods included Band Power (BP), Inter-trial Variance (IV) and Hilbert Transforms (HT). On the other hand, the dynamical behavior of the brain activities was well addressed by the graphical techniques like second order difference plot (SODP) and phase space reconstruction (PSR). The research was carried out on two different datasets one is obtained from the website of BCI competitions and another is recorded in-house. The electroencephalogram (EEG) signals of both normal and paralyzed subjects have been recorded by indigenously developed data acquisition system. Experimental results show that the proposed techniques effectively analyzed the dynamical behavior of brain activities and are superior to state-of-the-art techniques.