A Validation Employing Convolutional Neural Network for the Radiographic Detection of Absence or Presence of Teeth

Abstract

Dental radiography plays an important role in clinical diagnosis, treatment and making decisions. In recent years, efforts have been made on developing techniques to detect objects in images. The aim of this study was to detect the absence or presence of teeth using an effective convolutional neural network, which reduces calculation times and has success rates greater than 95%. A total of 8000 dental panoramic images were collected. Each image and each tooth was categorized, independently and manually, by two experts with more than three years of experience in general dentistry. The neural network used consists of two main layers: object detection and classification, which is the support of the previous one. A Matterport Mask RCNN was employed in the object detection. A ResNet (Atrous Convolution) was employed in the classification layer. The neural model achieved a total loss of 0.76% (accuracy of 99.24%). The architecture used in the present study returned an almost perfect accuracy in detecting teeth on images from different devices and different pathologies and ages.

A Convolutional Neural Network for Automatic Tooth Numbering in Panoramic Images

“…The neural network employed in this study was first constructed to automatically detect the presence or absence of a tooth with an accuracy of 99.24%, according to a previous author's manuscript [ 9 ]. Therefore, it was modified to add a new task which is tooth numbering employing FDI classification.…”

“…For this study, the 5,121 8-bits images employed in a previous published manuscript by the authors [ 9 ] were used to start the image database in the present study. A set of 2,230 correctly demarcated samples was obtained.…”

“…The same Matterport Mask RCNN employed in our previous study was employed [ 9 ] in the object detection ( Figure 2 ).…”

“…A ResNet101 was employed in the classification layer ( Figure 3 ). The classification layer was the same as the previous study [ 9 ] although to improve the automatic teeth numbering a new classification level was included (COCO).…”

“…The objective of this study was to modify the neural network used in a previous study by the authors [ 9 ], which obtained a precision of 99.24% in detecting the presence or absence of a tooth, to be used in the numbering of teeth in a panoramic image according to the Federation Dentaire Internationale (FDI) teeth numbering system.…”