Using magneto-inertial-measurement-units to track upper-limb movement during rehabilitation

Functional Electrical Stimulation (FES) can be used to support upper-limb rehabilitation after a stroke. A key aspect of FES control and also patient monitoring is the automatic tracking of upper-limb motion during intensive and functional practise of upper-limb tasks. To achieve this in a home environment, simple on-body sensors are required. A promising approach is to use Magnetic and Inertial Measurement Units (MIMUs), but they provide body-segment orientations rather than anatomical joint angles, the latter being more meaningful. To solve this problem the sensor orientation data must be interpreted anatomically, which requires that for each body-segment the orientation of its sensor coordinate frame is known with respect to its anatomical coordinate frame. Therefore, appropriate calibration must be performed to obtain the relationship between each sensor frame and its corresponding body-segment anatomical frame.While many papers have been published on anatomical calibration methods for MIMUs, there has been no comprehensive comparison of the alternative approaches to establish their relative merits. For FES supported upper-limb therapy, the need is for simple and fast donning and calibration, whilst achieving acceptable accuracy and repeatability with regards to the calculated joint kinematics. Therefore, the main objective of the PhD research was to undertake such a comparison and make recommendations for donning and calibration for the purposes of upper-limb FES.To address this problem the PhD work included:1. Undertaking a comprehensive and critical comparison of alternative anatomical calibration methods for MIMUs in terms of accuracy, speed, and simplicity.2. Finding the most appropriate anatomical calibration methods for use in upper-limb FES applications with stroke patients.3. Determining the best methods for processing MIMU outputs to provide anatomically meaningful upper-limb kinematic data.4. Experimentally assessing these methods against a gold standard (a VICON optical motion capture system).The results demonstrate that there is considerable variation between the alternative sensor defined anatomical frames and, hence, confirm the need for comprehensive comparisons. The comparisons reported in this thesis have led to tentative recommendations. Nevertheless, the methods reported are a sound foundation for future work to provide stronger recommendations, with more formal measures of confidence