Exploring the potential of machine learning models to predict nasal measurements through facial landmarks

Statement of problem. Information on predicting the measurements of the nose from selected facial landmarks to assist in maxillofacial prosthodontics is lacking. Purpose. The objective of this study was to identify the efficiency of machine learning models in predicting the length and width of the nose from selected facial landmarks. Material and methods. Two-dimensional frontal and lateral photographs were made of 100 men and 100 women. Different eye, nose, and ear landmarks were manually measured on the digital images. Various machine learning regression techniques were validated to confirm the accuracy of the length of the nose (LON), nasal bridge length (NBL), lateral alar width (LAW), and nasal tip protrusion (NTP). Results. The regression models used in this study to predict the width and length of the nose parameters demonstrated a robust predictive capability, as evidenced by the high coefficient of determination score obtained (greater than 0.95). This coefficient of determination suggested the implemented model was able to effectively capture the underlying patterns and relationships within the data, leading to enhanced efficiency in predicting the outcomes. SHapley Additive ExPlanations values demonstrated that for the men-only and women-only datasets, the measurement of the auricular projection was the most important predictor of the LAW, the ear length was the most significant contributor to the LON, and the angle between the nasal bridge length and the ear length was the most significant contributor to the NBL and NTP. For the combined datasets, the distance between the superior edge of the ear to the line measuring the medial canthus width contributed most to the LON and NBL, the ear length was the most significant contributor to LAW, and the angle between the nasal bridge length and the ear length was the most significant predictor of NTP. Conclusions. All selected algorithms provided precise width and length predictions for all data groups and were highly correlated with the actual value. The frontal images can be used to predict the LON and LAW, whereas the lateral images can be used to evaluate the NBL and NTP. (J Prosthet Dent 2025;133:1604-1613)

Journal

JOURNAL OF PROSTHETIC DENTISTRY

Year

2025

Volume

133

Issue

6

Pages

1604-1613

ISSN

0022-3913

eISSN

1097-6841

DOI

10.1016/j.prosdent.2024.12.012

WoS UID

WOS:001509185600032

Open via DOI