Design of a Robotic Rehabilitation System for Mild Cognitive Impairment Based on Computer Vision

Abstract

This paper develops a robotic cognitive rehabilitation therapy (CRT) system to assist patients with mild cognitive impairment (MCI) in block design test (BDT) rehabilitation training. This system bridges the treatment gap that occurs when one physician has several patients to attend to. One physician can setup the BDT training task and simultaneously monitor the training progress of several patients with MCI, which forms an effective one-to-many rehabilitation model. A target information acquisition method is designed to realize target detection and position extraction in automatic rehabilitation. Two graphic user interfaces (GUIs) are developed to provide intuitive control and immediate visual feedback. Different BDTs are selected from the benchmark by the physician in an integrated GUI (I-GUI) and are assigned to several patient GUIs (P-GUIs), respectively. During training, automatic visual assistance can be triggered by the help button and the patients can be guided in finding the target block. Additionally, a robotic arm could be engaged to further help with teaching so that patients can follow the instructions given by the P-GUI and imitate the demonstration given by the robot arm to finish the training task. This system converts traditional MCI rehabilitation into an automatic process, creating an effective model of BDT training for MCI rehabilitation.