Patient-tailored classification for a NIRS triggered hand rehabilitation robot

Abstract

Robotic neurorehabilitation that provides the support movement for the affected limb triggered by brain signal has a great potential to improve the recovery for post-stroke patients. We are studying a hand rehabilitation robotic system that a robotic hand orthosis is moved triggered by Near-Infrared Spectroscopy. In this paper, we propose a new method to classify the motion intention out of the NIRS signal. The classification accuracy that is an essential factor to extract the users' motion intension, was significantly improved by parameterizing the individual hemodynamic response.