Classifying Imaginations of Rhythmic Arm Movements in Two Planes from EEG

A brain-computer interface (BCI) can be used to control a limb neuroprosthesis with motor imaginations (MI) to restore limb functionality in paralyzed persons. However, existing BCIs lack a natural control and need a considerable amount of training time or may use invasively recorded brain signals. A new approach is the direct decoding of movements which has already been shown non-invasively for executed movements. In this work we show indirectly that algorithm principles used in decoding executed movements can also be applied when decoding imagined movements. Healthy subjects performed rhythmic arm movement imaginations in the transverse and sagittal plane. We were able to classify the correct movement plane with an average classification accuracy of 69 % considering only significant results. This shows that the classification of movement imaginations with the same hand in two different planes is possible.