Jonas Bianchi

Date of Award

9-26-2025

Department

Department of Orthodontics

First Advisor

Heesoo Oh

Abstract

Objectives: The primary aim of this study was to evaluate the accuracy and precision of a deep learning artificial intelligence (AI) model in reconstructing 3D facial geometry from a single 2D photograph. The study specifically measured the coordinate differences of human-placed facial landmarks on the AI-generated models against those placed on 3D models from a gold-standard scanner (Bellus3D). Secondary objectives included identifying dimensional variations (X, Y, Z) and systematic biases in the AI reconstructions. Methods: This was a retrospective comparative study using a sample of 47 anonymized patient records from the University of the Pacific database. For each patient, a 3D facial model was generated from a 2D photograph using the 3DDFA-V2 deep learning model. These AI-generated models were compared to reference models captured on the same day using a Bellus3D scanner. Two trained observers identified a set of facial landmarks on both the AI-generated and Bellus3D models. The coordinate data were normalized using Procrustes analysis to compare shape and landmark position independent of scale and orientation. The differences in landmark coordinates between the two model types were then statistically analyzed. Results: The analysis revealed that human landmark identification on the AI-generated models showed consistently lower variability compared to the gold-standard scans, suggesting that the smoother surfaces of the photo-reconstructed models may lead to more repeatable measurements. While the AI model demonstrated high fidelity, systematic errors were observed; for instance, a tendency to place certain landmarks (e.g., Me, Pog) in a more inferior position (negative Y-axis error). The greatest overall deviations (Euclidean error) were found in landmarks associated with the lateral and inferior borders of the face, such as Right Gonion (RGo), Trichion (Tr), and Left Gonion (LGo). Conclusions: The AI deep learning model is a promising low-cost tool for generating 3D facial models from 2D photographs with a high degree of precision in landmark placement. While the AI models facilitated more repeatable landmarking, they also exhibited some systematic biases. These findings have clinical implications for the use of photo-reconstructed models in diagnosis and treatment planning, highlighting areas where further refinement of the AI algorithm is needed.

Download

Share

COinS