Protocol design and testing to investigate Motor Imagery training using cues in different mediums: A pilot study

Sonal Santosh Baberwal, Luz Alejandra Magre, K. R. Sanjaya D. Gunawardhana, Tomás Ward, Shirley Coyle

Abstract: A Brain-Computer Interface (BCI) system converts brain activities into control signals that could be used to operate an interface such as an exoskeleton. One of the potential brain activities for BCI control is Motor Imagery (MI), i.e. imagination of limb movements. This type of brain activity is most prominent in the Alpha and Beta band. These signals require intensive training that could further be used as enhanced features for classification. Different studies highlight the importance of training by commands/cues presented in mediums such as a computer screen, audio or Virtual Reality (VR). However, as of now, there is little evidence regarding the use of this protocol for comparing the performance of these different training cues. This paper presents not only the design of the protocol but also reports a pilot study performed to evaluate MI signals by presenting cues in three different mediums, i.e. a computer screen, a VR headset, and speakers, studying the signals, extracting features and reporting the accuracy for the left/right classification. The results suggested a significant difference in the Event-Related Desynchronization Synchronization(ERDS) from MI signals of the participants for both alpha and beta bands when the cues were presented in the VR headset, along with the highest classification accuracy for this medium.