Developing a prototype using IMUs to do 24 hours measurement of knee joint mobility in children with Cerebral Palsy

Abstract: Cerebral Palsy (CP) is a disorder of movement that is caused before, during or soon after birth. CP affects movement and co-ordination caused by lack of muscle control by the brain [1], [2]. Stiff muscles, weak muscles, tremors and seizures as well as difficulties with swallowing are examples of symptoms in children with CP. Every year there are approximately 200 children born in Sweden that are affected by CP [3]. Children affected by CP must visit physicians or physiotherapists for an evaluation check up on their joints. Physicians use a goniometer to study the knee flexion/extension angle thus providing a plan for rehabilitation treatments. A challenge for the physiotherapists is that they are unable to measure the knee joint movement of children with CP for a longer period. There is a lack of research in the area of CP combined with long term measurement devices. This thesis project had a main purpose to create a prototype that was able to measure the knee joint angle in children with CP in their normal environment, for 24 hours. Two BNO055 inertial measurement units (IMUs) were used to obtain the knee joint angle by using quaternions, which are a number system in four dimensions. The final prototype was tested on a robot and human subject. The results concluded that BNO055 is a good IMU when the measurements are performed for a short time with slow or linear movements. For longer periods, however, the sensors lose their calibration and provide drifted results, hence it is not suited for doing measurements for 24 hours. Another upgraded and high cost sensor BNO080 was then replaced that provided more accurate results for knee joint angles over a longer period of motion. BNO080 also provided short drifting results over time, but not as bad as BNO055. Due to its many operational modes, it might be possible to overcome drifting errors. It was possible to develop a device for knee joint measurements with promising results for shorter periods, however, to continuously measure data for 24 hours on nonlinear human knee motions, a strong IMU with good autocalibration techniques is needed, and this is something that needs to be tested and studied further.