Hand-Movement Prediction Using LFP Data

Abstract

The last decade has seen a surge in the development of Brain-Machine Interfaces (BMI) as assistive neural devices for paralysis patients. Current BMI research typically involves a subject performing movements by controlling a robotic prosthesis. The neural signal that we consider for analysis is the Local Field Potential (LFP). The LFP is a low frequency neural signal recorded from intra-cortical electrodes, and has been recognized as one containing movement information. This thesis investigates hand-movement prediction using LFP data as input. In Chapter 1, we give an overview of Brain Machine Interfaces. In Chapter 2, we review the necessary concepts in time series analysis and pattern recognition. In the final chapter, we discuss classification accuracies when considering Summed power and Coherence as feature vectors.