Related Experiment Video
Updated: May 21, 2025
05:49
Author Spotlight: Deciphering Electrical Networks Behind Complex Brain Activities and Disorders
Published on: November 1, 2024
651
The accuracy of automated facial landmarking - a comparative study between Cliniface software and patch-based
Bodore Al-Baker1, Xiangyang Ju2, Peter Mossey3
1Orthodontic Department, Hamad Dental Centre, Hamad Medical corporation, Onaiza Service Road, P.O. Box 3050, Doha, Qatar.
European Journal of Orthodontics
|March 20, 2025
Summary
A new patch-based Convolutional Neural Network (CNN) algorithm demonstrates superior accuracy in facial landmark detection compared to Cliniface software. This CNN approach offers reliable precision for clinical evaluation of 3D facial images.
Area of Science:
- Medical Imaging
- Computer Vision
- Biomedical Engineering
Background:
- Automatic landmarking software aids 3D facial image analysis but often lacks clinical accuracy.
- Cliniface is an accessible open-access software for automatic facial landmarking with unverified clinical validity.
Purpose of the Study:
- To evaluate the accuracy of Cliniface software for facial landmark identification.
- To compare Cliniface's accuracy against a developed patch-based Convolutional Neural Network (CNN) algorithm.
Main Methods:
- The study analyzed 30 3D facial images with 20 anatomical landmarks.
- Manual digitization by an expert served as the ground truth.
- Cliniface and a patch-based CNN algorithm automatically detected landmarks.
- Partial Procrustes Analysis measured Euclidean distances for accuracy assessment.
Main Results:
- Cliniface software had an overall landmark localization error of 3.66 ± 1.53 mm, with significant discrepancies at landmarks like Subalar.
- The patch-based CNN algorithm achieved accuracy comparable to manual precision.
- Cliniface's inaccuracy was significantly higher than manual landmarking precision.
Conclusions:
- The patch-based CNN algorithm provides satisfactory accuracy for clinical evaluation of 3D facial images.
- Cliniface software's accuracy limitations restrict its clinical application for certain landmarks.
