A gait character analyzing system for osteoarthritis pre-diagnosis using RGB-D camera and supervised classifier

Abstract

The gait-related disease, such as osteoarthritis (OA) is a crippling disease which is the most prevalent form of arthritis in the elderly with an incidence rate of more than 50%. In today’s clinical diagnosing, the physicians always judge and record the gait of a patient qualitatively. Therefore, a cheap, easy-to-use gait analyzing system is important to achieve a quantified description and recording for both newly diagnosed patients and the follow-up patients. This study proposed an accurate gait analysis method by using RGB-D camera and supervised classifier. Firstly, we set up a gait assessment framework for OA patients using the RGB-D camera; design a joint data acquisition software to build a clinical setting. Secondly, the joint data of both patients and healthy controls from the sensor are analyzed to acquire fourteen quantitative gait parameters. Finally, a supervised SVM classifier is trained on the gait parameters of OA patients to help automatically diagnosis the gait anomalies. Experimental results demonstrated that gait parameters between OA patient and healthy controls shows significant different according to unpaired t-test. The average accuracy of the gait classification could reach 97%. Therefore, our study offers a scientific approach for quantitative, non-interactive and low-cost analysis of the gait, which can facilitate the diagnosis and treatment of the gait-related disease.