Related Experiment Video
Updated: Oct 31, 2025
Author Spotlight: Insights into Visual Cortex Research Through Wide-View fMRI Mapping
Published on: December 8, 2023
Tic Detection in Tourette Syndrome Patients Based on Unsupervised Visual Feature Learning
Junya Wu1, Tianshu Zhou2, Yufan Guo3
1Engineering Research Center of EMR and Intelligent Expert System, Ministry of Education, Key Laboratory for Biomedical Engineering of Ministry of Education, College of Biomedical Engineering and Instrument Science, Zhejiang University, Hangzhou 310027, China.
This study introduces an automated deep learning method for detecting tic movements from videos, aiding in tic disorder diagnosis. The model shows promising accuracy, potentially improving clinical evaluation and treatment monitoring.
Area of Science:
- Medical technology
- Artificial intelligence in healthcare
- Neurology
Background:
- Clinical diagnosis of tic disorders is complex, relying on time-consuming manual observation and evaluation.
- Current assessment scales for tic disorders require manual completion, highlighting a need for automation.
Purpose of the Study:
- To develop an automatic method for detecting tic movements using deep learning to assist in the diagnosis and evaluation of tic disorders.
- To leverage clinical video data for training a model that can differentiate between tic movements and non-tic behaviors.
Main Methods:
- A deep learning architecture combining unsupervised and supervised learning was proposed.
- The model was trained on clinical video data using leave-one-subject-out cross-validation for binary and multiclass classification.
- Feature learning from unsupervised stages was visualized to assess tic distinguishability.
Main Results:
- The model achieved average recognition precisions of 86.33% (binary) and 86.26% (multiclass).
- Average recalls were 77.07% (binary) and 78.78% (multiclass).
- Visualizations confirmed the model's ability to distinguish between different types of tics and non-tic movements.
Conclusions:
- The proposed deep learning model demonstrates potential for automatic tic movement detection.
- This technology could serve as an auxiliary tool for clinical diagnosis of tic disorders.
- The approach shows promise for evaluating treatment effects in patients with tic disorders.
