Precision Assessment of Facial Asymmetry Using 3D Imaging and Artificial Intelligence.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Orthodontics

Background:

• Growing interest in AI for enhancing diagnostic precision and efficiency.
• Facial asymmetry assessment is crucial in orthodontics and other fields.
• 3D facial imaging offers detailed anatomical data.

Purpose of the Study:

• To assess the precision of an AI-based method for facial asymmetry analysis.
• To compare AI-driven landmark identification and asymmetry index calculation with manual methods.
• To evaluate the reliability of manual measurements in 3D facial imaging.

Main Methods:

• Analysis of 3D facial images from 130 patients using Vectra® M3 system.
• Manual identification of seven bilateral facial landmarks and calculation of asymmetry index.
• Development of an AI program for automated landmark identification and asymmetry index calculation.
• Assessment of manual measurement reliability using intraclass correlation coefficients (ICC).

Main Results:

• Manual measurements showed moderate to excellent reliability (ICC 0.62-0.99 within, 0.72-0.96 between raters).
• Agreement between manual and AI methods for asymmetry index in five landmarks.
• Statistically significant differences in asymmetry index for alare (p=0.0056) and cheilion (p=0.0081) between methods.

Conclusions:

• AI-based facial asymmetry analysis using 3D images offers efficient and comparable precision.
• Observed discrepancies between AI and manual methods require further software improvement and training.
• This AI approach holds potential for advancements in orthodontic research and clinical practice.