Abstract
This thesis describes the design and development of a prototype, rhythm-based game system for upper limb rehabilitation after stroke with focus on motion analysis and data analytics to provide feedback on clinically relevant movements to optimise motor control. Growing evidence supports the benefits of game technology for rehabilitation after stroke. However, as this technology will typically be used at home with little supervision, it is crucial that systems are in place to detect if exercises are performed safely and effectively. Despite this, few systems tailor the game activity to the individual needs of the stroke survivor, integrate motion tracking algorithms to detect harmful activity, or provide a variety of quantitative metrics that are clinically relevant and easy to understand which can be used to assess the stroke survivor’s progress.To address this gap in research, a prototype motion capture game system has been developed which incorporates Rhythmic Auditory Cueing (the practice of synchronising rehabilitation movements with a periodic rhythm), using preferred music as the rhythmic stimuli to facilitate reaching movements of the affected arm. Within the game system, a variety of calibration processes have been developed to assess the functional ability of the user and constrain the game activity to appropriate and safe ranges. To detect potentially harmful activity when engaging with the game system, an algorithm has been developed which analyses reaching motions in real-time and detects compensatory movements via a low-cost depth camera. The algorithm observes the motion of the user and segments cyclical reaching motions into discrete actions and four movement types which describe and quantify the motion of the upper limb and trunk. These movement types are trunk forward leaning, trunk rotation, shoulder rotation, and forward reach. To explore the estimated error within the system, the data collected by the game camera was compared to a marker-based motion capture system. Results show the game system camera operates with lower accuracy when the user has restricted range of motion, with a mean error of 7.8% reported across all movement types.
To assist health professionals in their work and address a further gap in research, an additional software tool has been developed which performs further analysis on the kinematic data output by the game system. The software graphically displays a variety of quantitative metrics related to motor function which allow health professionals to monitor the performance and progress of the stroke survivor. The metrics available within the software are compensatory movements, arm extension, movement speed, smoothness, and velocity. The software was appraised by experts in stroke and rehabilitation, with results showing that measures which quantify the use of the upper limb (and any compensatory movement which hinder this use) are seen as valuable, the terminology employed throughout the software must be self-explanatory and non-technical, and graphical representations of data are seen as beneficial.
| Date of Award | 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Don Knox (Supervisor) & Frederike van Wijck (Supervisor) |