Application of deep learning to classify skeletal growth phase on 3D radiographs

Abstract Cervical vertebral maturation (CVM) is widely used to evaluate growth potential in the field of orthodontics. The aim of this study is to develop an artificial intelligence (AI) algorithm to automatically predict the CVM stages in terms of growth phases using the cone-beam computed tomographic (CBCT) images. A total of 30,016 slices obtained from 56 patients with the age range of 7-16 years were included in the dataset. After cropping the region of interest (ROI), a convolutional neural network (CNN) was built to classify the slices based on the presence of a good vision of vertebrae for classification of the growth stages. The output was used to train another model capable of categorizing the slices into phases of growth, which were defined as Phase I (prepubertal, CVM stages 1 and 2), phase II (circumpubertal, CVM stage 3), and phase III (postpubertal, CVM stages 4, 5, and 6). After training the model, 88 unused images belonging to 3 phases were used to evaluate the performance of the model using multi-class classification metrics. The average classification accuracy of the first and second CNN-based deep learning models were 96.06% and 95.79%, respectively on the validation dataset. The multi-class classification metrics applied to the new testing dataset also showed an overall accuracy of 84% for predicting the growth phase. Moreover, phase I ranked the highest accuracy in terms of F1 score (87%), followed by phase II (83%), and phase III (80%) on new images. Our proposed models could automatically detect the C2-C4 vertebrae required for CVM staging and accurately classify slices into 3 growth phases without the need for annotating the shape and configuration of vertebrae. This will result in developing a fully automatic and less complex system with reasonable performance, comparable to expert practitioners. Author Summary The skeletal age of orthodontic patients is a critical factor in planning the proper orthodontic treatment. Thus, an accurate assessment of the growth stage can result in better treatment outcomes and reduced treatment time. Traditionally, 2-D cephalometric radiographs obtained during the orthodontic examination were used for estimating the skeletal age using the three cervical vertebrae. However, this method was subjective and prone to errors as different orthodontists could interpret the features differently. Moreover, 2-D images provide only limited information as they only capture two dimensions and involve superimpositions of neighbour structures. In the present study, machine learning models are applied to 3-D cephalometric images to predict the growth stage of patients by analyzing the shape and pattern of cervical vertebrae. This method has the potential to improve treatment outcomes and reduce the treatment time for orthodontic patients. Additionally, it can contribute to the development of more personalized treatment plans and advance our understanding of the growth and development of the craniofacial complex.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.