Decoding Action Planning of three-dimensional Movements Using Electrocorticographic signals

Abstract

A brain machine interface (BMI) is designed to convert raw neural signals into motor commands and reproduces the movements of the body with a neuroprosthetic device. However, little is known about the motor planning period which entails the intention and planning of the movement. To analyze the planning data, we asked subjects to perform a reach-and-grasp motion. We decoded the planning (-2 to 0 s from the movement onset) and execution (0 to 2 s from the movement onset) data. The decoding performance of the planning period was similar (Pearson’$\mathrm{sp}\lt 0.05)$ to the execution period. Additionally, we calculated the connectivity of the planning and execution period. We compared the connectivity data between motor planning and movement execution (ME). Our hypothesis is that the neural mechanism of the motor planning signal is similar to the execution signal because movement uses movement planning and intention. Connectivity was analyzed to compare the network pattern of the planning period and execution period. The results showed that the planning period of the ME signal is significantly similar (Pearson’$\mathrm{sp}\lt 0.05)$ to the connectivity matrix of the execution period of the ME signal. The results suggest that the neural network during motor planning is similar to the actual execution and thus important in the control of BMIs.