Auditory imagery classification with a non-invasive BCI

Methods such as EEG and fMRI, to name a few, have allowed a clearer view regarding brain activity and function. Noninvasively, more details can be found regarding the brain areas that get stimulated while imagining sounds. In motor imagery this is a well-known and well-defined topic, while in auditory imagery is not. It is possible to try to establish key differences between stimuli presentation (listening), imagery periods and silence, to verify and characterize their respective brain activity. Experiments were performed over a group of 16 individuals, using white noise as stimulus signal and open software for a noninvasive BCI, typical scenarios of motor intention classification were used for auditory imagery analysis. Although, is possible to achieve beyond a 90% in a two discrimination class offline, the 70.999% average percentage of success in classifying between auditory imagery and silence, indicates that it may lead to a significant comparability in future work.