Automatic Detection of Facial Tics in Gilles de la Tourette Syndrome Patients

Rijn, F.W. van

Automatic Detection of Facial Tics in Gilles de la Tourette Syndrome Patients

Files

Rijn van_f_BA_Th 2014.pdf (2.71 MB)

Authors

Rijn, F.W. van

Issue Date

2014-01-30

Language

en

URI

http://theses.ubn.ru.nl/handle/123456789/145

Abstract

This project pursues automatic facial tic detection in video recordings of Gilles de la Tourette syn- drome patients. Current diagnostic methods using a severity measurement scales and tic counts require huge e ort from the therapist and the patient. By automating these diagnostic methods, this would alle- viate these e orts. A study by Chappell et al. shows that manually counting tics in videotape recordings is feasible[9]. Only few studies used arti cial intelligence to automate the process of tic detection. A study by M. Bernabei is the most prominent one and shows that automatic tic detection with triaxial accelerometers is possible[3]. In the current project, distinctive features used by experts have been explored for their use in a system using computer vision and machine learning techniques. An implementation for this system is made and the performance is benchmarked. This project explored whether this approach was feasible and yields recommendations for further improvements. Finally, results and guidelines for the design of a system that uses automated tic detection for psychotherapeutic feedback and exercises are given.