Learning Cephalometric Landmarks for Diagnostic Features Using Regression Trees

Lateral cephalograms provide important information regarding dental, skeletal, and soft-tissue parameters that are critical for orthodontic diagnosis and treatment planning. Several machine learning methods have previously been used for the automated localization of diagnostically relevant landmarks...

Full description

Saved in:
Bibliographic Details
Published inBioengineering (Basel) Vol. 9; no. 11; p. 617
Main Authors Suhail, Sameera, Harris, Kayla, Sinha, Gaurav, Schmidt, Maayan, Durgekar, Sujala, Mehta, Shivam, Upadhyay, Madhur
Format Journal Article
LanguageEnglish
Published Basel MDPI AG 27.10.2022
MDPI
Subjects
Accuracy
Algorithms
anatomical landmarks
Automation
Bioengineering
cephalograms
Datasets
Labeling
Localization
Machine learning
Medical imaging
Methods
Orthodontics
Radiography
Regression analysis
regression trees
Trees
X-rays
United States
Online AccessGet full text

Cover

More Information
Summary:Lateral cephalograms provide important information regarding dental, skeletal, and soft-tissue parameters that are critical for orthodontic diagnosis and treatment planning. Several machine learning methods have previously been used for the automated localization of diagnostically relevant landmarks on lateral cephalograms. In this study, we applied an ensemble of regression trees to solve this problem. We found that despite the limited size of manually labeled images, we can improve the performance of landmark detection by augmenting the training set using a battery of simple image transforms. We further demonstrated the calculation of second-order features encoding the relative locations of landmarks, which are diagnostically more important than individual landmarks.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:2306-5354
2306-5354
DOI:10.3390/bioengineering9110617