Fusion Methods in Motor Imagery Classification

Motor Imagery (MI) based Brain-Computer Interface system (BCI) has various applications including prosthesis control, virtual reality control, motor rehabilitation, etc. Yet, such systems still face many challenges such as low accuracy in MI classification. There has been a lot of research on improving motor imagery performance, but most of them focus on studying single signal or single technique. In this paper, we explore the feasibility of improving the accuracy using fusion methods. First, we will explain some terminology regarding to motor imagery and brain computer interface. Then, we will discuss several categories of fusion according to the level at which the fusion is conducted. Finally, we discuss different situations and different fusion methods specifically. This review summarizes typical technology and state-of-the-art progress in the fusion methods of motor imagery classification over the last decades.