A Convolutional Neural Network for Automatic Tooth Numbering in Panoramic Images

Abstract

Analysis of dental radiographs and images is an important and common part of the diagnostic process in daily clinical practice. During the diagnostic process, the dentist must interpret, among others, tooth numbering. This study is aimed at proposing a convolutional neural network (CNN) that performs this task automatically for panoramic radiographs. A total of 8,000 panoramic images were categorized by two experts with more than three years of experience in general dentistry. The neural network consists of two main layers: object detection and classification, which is the support of the previous one and a transfer learning to improve computing time and precision. A Matterport Mask RCNN was employed in the object detection. A ResNet101 was employed in the classification layer. The neural model achieved a total loss of 6.17% (accuracy of 93.83%). The architecture of the model achieved an accuracy of 99.24% in tooth detection and 93.83% in numbering teeth with different oral health conditions.

A Comprehensive Exploration of Neural Networks for Forensic Analysis of Adult Single Tooth X-Ray Images

“…Additionally, while not going into an exhaustive analysis, they have some insight into the impact of tooth alterations, claiming that "the network is capable of correctly numbering teeth that contain metal parts, or any other treatment performed on it such as filled teeth, but in the case of the prosthetic crown, it detects a single tooth." [93] Our model achieves only an accuracy of 87.24% on healthy, unaltered teeth for the 32-class case and an overall accuracy of 83.74%. However, as [93] uses the 8class approach, which is then post-processed into the 32class case, we believe that comparison to our 8-class results is more appropriate.…”

“…[93] Our model achieves only an accuracy of 87.24% on healthy, unaltered teeth for the 32-class case and an overall accuracy of 83.74%. However, as [93] uses the 8class approach, which is then post-processed into the 32class case, we believe that comparison to our 8-class results is more appropriate. In that case, we achieve an accuracy of 95.53% on healthy, unaltered teeth and an overall accuracy of 92.40%.…”

“…Sample size Classes Accuracy Oktay et al [53] 100 OPGs 4 types 92.84% Keerthana et al [71] 200 tooth x-rays 4 types 92.54% Chen et al [72] 1250 dental periapical films 32 types 71.5% to 91.7% Prados-Privado et al [93] consists of 1250 dental periapical films. The accuracy of their model varies between 71.5% to 91.7%.…”

“…While relatively close in performance, we believe that our model performs better due to sample size, model capacity, and the solitary focus on tooth type determination. Prados-Privado et al [93] determine the tooth type on panoramic dental x-ray images. They do not use any neighborhood-based postprocessing of results, but they use symmetries of teeth to simplify their problem from the 32-class approach to the 8-class approach.…”

Convolutional neural networks for automated tooth numbering on panoramic radiographs: A scoping review

“…Ten studies utilized a distinct test dataset. In 2 studies, 26 27 the total number of images was the only information provided, with no details regarding the partitioning of the dataset. In 6 of the 10 studies with a distinct test set, 24 25 28 30 31 32 data were divided into 3 datasets (training, validation, and testing), while 4 studies 21 22 23 29 reported division into 2 datasets (training and testing).…”