POST-STROKE REHABILITATION: STICK EXERCISE MONITORING USING KALMAN FILTER

Abstract

Post-stroke rehabilitation is a necessary step to improve the function of motor and muscle of the patients. However, it is difficult for the patients to have one-to-one session with the physiotherapist at the rehabilitation center due to the constraint of money, location and time. Thus, there are many existing inventions that provide home-based physiotherapy monitoring to facilitate the patients in performing the rehabilitation exercises to restore the body functions and mechanical skills. This paper presents a novel approach to monitor the stick exercises, in which the system will count the number of exercise cycle performed by the patient for a certain period of time. By using Kalman filter approach, two inputs are observed based on the color of patient’s cloth and length of the stick. The centroid coordinate obtained from the Kalman filter output is then used to map the motion graph. Two limits, upper and lower boundaries are set such that a complete cycle of exercise is confirmed if the patient passes both of the boundaries. Those limits are determined based on the ratio of human body proportion. Eight simulated videos of the stick exercise are used to validate the proposed method. The results show that the best performance with 100% correctly count is obtained from video 3 while the worst performance is taken from video 2 and 8. The system can be further improved by incorporating different types of shoulder exercises for post-stroke patient