Influence of growth structures and fixed appliances on automated cephalometric landmark recognition with a customized convolutional neural network

Abstract

Background One of the main uses of artificial intelligence in the field of orthodontics is automated cephalometric analysis. Aim of the present study was to evaluate whether developmental stages of a dentition, fixed orthodontic appliances or other dental appliances may affect detection of cephalometric landmarks. Methods For the purposes of this study a Convolutional Neural Network (CNN) for automated detection of cephalometric landmarks was developed. The model was trained on 430 cephalometric radiographs and its performance was then tested on 460 new radiographs. The accuracy of landmark detection in patients with permanent dentition was compared with that in patients with mixed dentition. Furthermore, the influence of fixed orthodontic appliances and orthodontic brackets and/or bands was investigated only in patients with permanent dentition. A t-test was performed to evaluate the mean radial errors (MREs) against the corresponding SDs for each landmark in the two categories, of which the significance was set at p < 0.05. Results The study showed significant differences in the recognition accuracy of the Ap-Inferior point and the Is-Superior point between patients with permanent dentition and mixed dentition, and no significant differences in the recognition process between patients without fixed orthodontic appliances and patients with orthodontic brackets and/or bands and other fixed orthodontic appliances. Conclusions The results indicated that growth structures and developmental stages of a dentition had an impact on the performance of the customized CNN model by dental cephalometric landmarks. Fixed orthodontic appliances such as brackets, bands, and other fixed orthodontic appliances, had no significant effect on the performance of the CNN model.

Application of Artificial Intelligence (AI) in a Cephalometric Analysis: A Narrative Review

“…Table 1 presents the summarised studies on the application of AI in cephalometric analysis. In total, 23 articles were included based on both AI algorithms designed by their authors for the purpose of a specific study [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ] and web-based software available on search engines and mobile applications [ 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 ]. The studies focused on comparing the reliability of AI algorithms in localising cephalometric landmarks on lateral cephalometric radiographs with the manual tracing of these points; differences between various algorithms were also examined [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 ].…”

“…The lowest accuracy was for the point Gonion, with 48.3% for the 2 mm error. Recently, Popova et al confirmed that the presence of orthodontic appliances did not significantly influence the performance of CNN-based open-source models, such as the Python programming language [ 25 ].…”

“…Recently, Popova et al assessed the influence of growth structures such as tooth buds and the presence of fixed orthodontic appliances on the accuracy of a customised CNN model for the automatic detection of cephalometric landmarks [ 25 ]. Sixteen skeletal and dental cephalometric landmarks were included in the analysis.…”

Prediction of Pubertal Mandibular Growth in Males with Class II Malocclusion by Utilizing Machine Learning

“…Within AI, ML utilizes a set of inputs and outputs to create an algorithm to process the data and correctly predict the output [ 19 ]. AI and ML have been utilized for several tasks in orthodontics, such as for automated cephalometric analyses [ 20 , 21 , 22 , 23 , 24 ], predicting extraction vs. non-extraction treatment decisions [ 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 ], predicting orthodontic extraction patterns [ 34 ], determining the need for surgery in Class III patients [ 35 ], and growth assessment [ 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 ]. However, little research has been conducted on the use of AI to predict mandibular growth.…”