Neuro-Fuzzy based motor imagery classification for a four class brain machine interface

Abstract

Brain Machine Interface (BMI) provides a digital link between the brain and a device such as a computer, robot or wheelchair. This paper presents a BMI design using Neuro-Fuzzy classifiers for controlling a wheelchair using EEG signals. EEG signals during motor imagery (MI) of left and right hand movements are recorded noninvasively at the sensorimotor cortex. Four mental task signals are analyzed and classified to design a four class BMI. The proposed classifier has an average classification performance of 97%.