Poster: Kinect-based automatic scoring system for spasmodic torticollis

摘要

Displacement and shaking of the human body are essential measures for the scoring of movement disorder. The aim of our research is to increase the accuracy and reproducibility of the scoring for spasmodic torticollis, a movement disorders that is characterized by poor head-posture. Although there are conventional scoring methods for torticollis severity, such as the Tsui Scale or TWSTRS, the results obtained are sometimes inaccurate or non-reproducible because these tests are performed manually. To capture the posture of patients' heads and bodies, we used Kinect, a three-dimensional depth sensor, and tracked their faces and bodies in real time. Using obtained data, the system automatically calculates TWSTRS without any special knowledge and training. An experiment comparing the accuracy of the system with medical specialist suggests that our system is sufficiently accurate.