Related Experiment Video
Updated: Aug 13, 2025
10:28
Dynamic Digital Biomarkers of Motor and Cognitive Function in Parkinson's Disease
Published on: July 24, 2019
15.3K
DeepSmile: Anomaly Detection Software for Facial Movement Assessment
Eder A Rodríguez Martínez1,2, Olga Polezhaeva1,3, Félix Marcellin1,2
1UR 7516 Laboratory CHIMERE, University of Picardie Jules Verne, 80039 Amiens, France.
Diagnostics (Basel, Switzerland)
|January 21, 2023
Summary
A new deep learning model accurately assesses facial palsy by detecting anomalies in smiles. This AI tool aids clinicians in diagnosing and managing facial movement disorders.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neurology
Background:
- Facial movements are vital for human communication and social interaction.
- Accurate facial movement analysis is crucial for diagnosing and managing facial palsy.
- Current assessment methods for facial palsy are often inaccurate, subjective, or rely on static data.
Purpose of the Study:
- To implement a deep learning algorithm for objective and accurate assessment of facial movements, specifically during smiling.
- To address the limitations of existing clinical methods for evaluating facial palsy.
Main Methods:
- A deep learning model was developed and trained on a dataset of healthy smiles using an anomaly detection strategy.
- The model computes the degree of anomaly by comparing a person's smile to a "healthy" smile profile.
- A graphical user interface (GUI) was created for practical clinical application.
Main Results:
- The deep learning model successfully identified a high degree of anomaly in the smiles of patients with facial palsy.
- The developed GUI demonstrated practical utility for routine clinical use.
Conclusions:
- The study presents a novel deep learning model for assessing facial movements.
- This AI-powered tool, implemented on open-source software, offers a promising solution to aid clinicians in facial palsy assessment.
