Classification of Forearm Movements by Using Movement Related Cortical Potentials

Brain Computer Interfaces (BCIs) are used together with upper extremity prostheses to gain mobility skills to people who have lost their motor control abilities. In BCIbased prosthetic and rehabilitation systems, the degree of freedom determines the number of variables that can be controlled. Increasing the degree of freedom is possible by increasing the number of movements that can be resolved in the same limb. Movement Related Cortical Potentials (MRCPs) are low-frequency amplitude shifts that begin to occur prior to voluntary movement. In this study, it is aimed to classify pronation and supination movements in the same forearm using MRCPs. For this purpose, it is suggested to use Higuchi's fractal dimensions and k-nearest neighbor methods. In this study, the effect of post-stimulus time interval and window size on classification accuracy was also investigated. The proposed method was validated on a dataset of attempted hand and arm movements whose participants were spinal cord injury patients. According to the results, pronation movement was determined with an accuracy of 70.09% and supination movement was determined with an accuracy of 69.22%. The highest classification accuracy was obtained using the 3s post-stimulus time interval and a 100 ms window size.