Convolutional Neural Network-Based Deep Learning Methods for Skeletal Growth Prediction in Dental Patients

Area of Science:

• Dentistry
• Radiology
• Artificial Intelligence

Background:

• Accurate skeletal growth assessment is crucial for orthodontic treatment planning.
• Traditional methods for evaluating skeletal maturation can be time-consuming and subjective.
• Deep learning offers potential for automated and precise analysis of dental radiographs.

Purpose of the Study:

• To predict skeletal growth maturation using deep learning (convolutional neural network - CNN) with cervical vertebral maturation (CVM) and lower 2nd molar calcification.
• To evaluate the accuracy of CNN-based multiclass classification for detecting skeletal maturation stages from orthopantomography (OPG).
• To compare the predictive accuracy between CVM and 2nd molar calcification for skeletal maturation assessment.

Main Methods:

• Utilized a dataset of 1200 cephalometric radiographs and 1200 OPGs.
• Employed CNN-based deep learning with multiclass classification to analyze images.
• Identified cervical vertebral maturation index (CVMI) and estimated chronological age from 2nd molar calcification levels.

Main Results:

• The CNN model achieved high accuracy in predicting skeletal maturation stages and gender.
• Cervical vertebral maturation showed 98% accuracy in males.
• 2nd molar calcification demonstrated high accuracy in females, proving reliable for growth assessment using OPGs.

Conclusions:

• CNN multiclass classification is an accurate method for detecting skeletal maturation from both CVM and lower 2nd molar calcification.
• Lower 2nd molar calcification is a reliable indicator of skeletal growth level.
• Standard orthopantomography (OPG) is sufficient for assessing skeletal maturation using this deep learning approach.