Loading...

Development of Upper-Limb Motion Performance Indices Used in Home-Based Rehabilitation Systems

Fakhar, Maliheh | 2013

463 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 44960 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Behzadipour, Saeed
  7. Abstract:
  8. In recent years, the national health systems have supported the idea of home-based rehabilitation, since receiving cares at hospital environments is too expensive both in money and time. This method reduces the patients’ commuting to the clinics. Moreover, it enables them to practice the prescribed movements at any time.Most of the existing systems compute the patient’s motion performance in a zero-one manner which does not distinguish any small progress in his condition. Thus, a non-task based method which can assess precisely the patient’s level is yet missing. This project aims to develop a set of motion performance indices which can judge the upper limb motions in a home-based rehabilitation system.To this goal, the Kinect sensor has been chosen as a marker-free motion tracking system with an affordable cost. Besides, in order to eliminate the noise from measured data and obtaining other kinematic parameters, a Kalman filter has been designed and its performance in simulation and experiment has been tested.In the next step, a database has been created using the motion data of a set of 6 healthy and 49 post-stroke cases. Here, each individual was asked to perform horizontal, vertical and diagonal movements with both limbs. The motion data were recorded using the Kinect sensor and the test session was filmed. Then, the cases were scored by a physiotherapist and an occupational therapist. They were asked to score their motion performances with integers from 1 to 5.By fully studying the existing motion performance indices in the literature, the kinematic parameters were chosen. Seven different indices were recognized and formulated. These indices were normalized between the right and the left limbs. Then, their cross correlations with the scores from the therapists were calculated. From the results, three indices were selected as the bests. These are those of shoulder movements, elbow flexion angle and the total number of jerk oscillations. At the end, a RBF neural network classifier was trained to map the input performance indices to the output scores. It could be said that this system can efficiently replace the role of physician judgment in a home-based rehabilitation system
  9. Keywords:
  10. Motion Performance ; Non-Task-based Index ; Kinect Sensor ; Home-Based Rehabilitation System ; Stroke

 Digital Object List

 Bookmark

No TOC