Towards the control of a cursor on the screen using a 4-class mi based BCI

Brain-computer interfaces (BCIs) offer an alternative method of communication between the subject and the external environment, without relying on conventional communication channels. The ability to modulate its own electroencephalogram EEG signal during the imagination of specific movements can enable the actuation of specific movements to external devices, thus repairing or reinforcing some abilities that may have been lost. Herein, we propose two possible setups for a four-class MI BCI, which aim to test the feasibility of these endogenous BCIs to accurately control a cursor on the screen. To this end, an initial calibration phase has been conducted for each of the 12 healthy subjects to create two initial decoders, for the X and Y axes respectively; after this phase, three online test runs for each axis have been conducted to test the efficacy of these decoders, Subjects have been allowed to pass to the control of the cursor in 2D only if they reached at least 70% accuracy on each axis singularly. Those who passed to this final stage proved that a good control may be possible with adequate time to train, stating that a four-class MI BCI is a feasible, although at its beginnings, system able to provide an alternative communication channel.