U-Net based landmark detection of facial structures in cephalometric images

View at publisher Summarise with VIRAT
Authors: G. Subramanian, Archita Akilan, Ramesh Muthusamy, Suresh Chinnathampy Manickam, Jeyalakshmi Vijayarajan, Bharathi Kumarasekar
Cited by: 0
References: 13

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Cephalometric analysis plays a vital role for diagnosing jaw and teeth misalignment. Traditionally,Itrelies on manually identifying key landmarks on X-rays which is a time-consuming and subjective process. This work explores the potential of deep learning to revolutionize this field by automating landmark detection using a U-Net architecture. This powerful model, trained on a set of 19 critical anatomical landmarks, excels at extracting specific features from complex medical images. Once trained, the U-Net model can analyze new X-rays, automatically identifying the same landmarks with its advanced image segmentation capabilities. This innovative approach, further enhanced by a multichannel heatmap regression technique, tackles the limitations of manual methods by aiming for both increased efficiency and improved accuracy. By automating landmark detection with high accuracy and speed, this research holds significant promise. It could streamline the analysis of cephalometric X-rays, leading to improved clinical decision-making by dentists and ultimately, contributing to better patient care in dental and skeletal diagnostics.

Fields of Study & Subjects

Dental Radiography and Imaging Orthodontics and Dentofacial Orthopedics Medical Image Segmentation Techniques

Publication details

Journal / source
AIP conference proceedings
Volume
3341
Pages
020061–020061
Publication date
1 January 2026
Language
EN