HomeRehab : development of a supportive tool for home rehabilitation based on a depth camera

Abstract

Medical physical rehabilitation or physiotherapy is the most universally adopted treatment strategy for reducing human motor deficiencies caused by injuries, diseases, stroke, infarctions or old age. Physical rehabilitation is accomplished through performing precisely determined rehabilitation exercises provided by therapists, learned and performed by the patients during therapy sessions. However, due to the fact that the number of people in need of rehabilitation, as well as the medical costs are constantly increasing, the number of therapy sessions per patient is limited. This necessitates the patients to additionally perform the exercises at home, where no feedback of a therapist regarding the correctness of the exercise execution is available. The aim of this work is to evaluate the suitability of low-cost depth-sensing devices for home-based computer-assisted rehabilitation. In the course of this master's thesis, a software tool aimed on assisting the patients while performing their required rehabilitation exercises at home has been developed and a first prototype application has been implemented and evaluated. The system is based on a low-cost depth-sensing camera for tracking the patients-motions. With the information provided by this depth sensor, a basic skeletal representation of the patients-body is generated and used for calculating aberrations of the patients-motions from predetermined reference exercises provided by therapists. Through this approach, the application is able to provide the patients with the necessary feedback to perform their exercises at home correctly and further is designed to motivate the patients for accurate and frequent exercise execution. The application can be used with arbitrary rehabilitation exercises and each exercise can be adjusted according to the patients-current condition by defining specific exercise parameters. The evaluation of the application revealed promising results in regard to the performance of distinguishing between correct and incorrect exercise executions and in regard to comprehensibility, usability and motivation.