Methodology for Selection of Ideal Sound Stimulation Frequencies for BMIs Applications Based on the Oddball Auditory Paradigm

Individuals with locked-in syndrome face several problems of communication and autonomy once they are trapped inside their own bodies. They can benefit from a Brain-Machine Interface system that allows communication and interaction with the world. Event-related potentials (ERPs), especially the P300 component, are commonly used as control signal in these cases. However, ERPs and also the BMIs system performance can be influenced by several features related to the biological factors and the stimulation. This work evaluates and select the ideal stimulation sound frequency associated with a more significant P300 potential in an oddball paradigm aiming at the best performance of BMIs. The experimental procedure presents five mono target frequencies at the two channels of the headphone. The amplitude and latency of P300 was analyzed and used as attributes to a SVM classifier to distinguish the two stimuli (target and standard) and select the best frequency associated with a higher accuracy. It was identified different individual cortical responses and frequencies that lead to more significant P300 component and better classification performance. The proposed method can be used to evaluate the cortical response and select a set of parameters associated with a more adapted, comfortable and efficient BMI for the final user, ensuring greater usability and satisfaction.