This project is focused on movement analysis of Parkinson’s disease patients. Patients with Parkinson’s suffer from severe movement pattern disturbances. Symptoms range from slow movements to rigidity and postural instability. Our line of research aims to develop non-obtrusive technologies to monitor these complex movement impairments, focusing on detecting and analyzing standing-up speed, walking speed and trunk angles. The analysis will be done on depth images obtained from a Kinect v2 camera.
Existing movement monitoring systems fall into two general categories. The first category contains sensing systems based on accelerometers. These are expensive and intrusive systems with multiple body straps and sticky patches that tend to disturb natural walking awareness. Video based systems are also used, but the most advanced of them still use markers worn on the body. While these are minimally intrusive devices, they are mostly designed for laboratory use with specialized personnel,
We are developing a non-intrusive monitoring system that relies on depth images. The system is based on a MS Kinect v2 device that actively projects its own time of flight infrared beams to capture depth images of the patient.
The main task of the student will be to find and analyze patterns in a dataset of depth camera recordings of Parkinson’s patients. Kinect v2 uses a proprietary segmentation algorithm developed by Microsoft and which is based on decision forests. This is classification algorithm based on learning methods that extracts the body silhouette from the background. Based on the silhouette, a skeleton is determined which is further used to assess movement parameters. This project aims at developing new methods walking movement analysis from depth images. Parameters may include walking speed, cadence, arm swing, and probably several more. The results will be evaluated by comparing them to ground-truth measurements from real world recordings. The project requires that the student has knowledge of C# and Matlab, and is keen to further develop this knowledge.
The research will be performed in the VCA group as a full-time project. Depth camera data and initial analysis software are available and will be explained and provided to the student. There will be regular progress meetings during which the student will present his intermediate results.