Proposing a CNN Method for Primary and Permanent Tooth Detection and Enumeration on Pediatric Dental Radiographs

Abstract

In this paper, we aimed to evaluate the performance of a deep learning system for automated tooth detection and numbering on pediatric panoramic radiographs. Study Design: YOLO V4, a CNN (Convolutional Neural Networks) based object detection model was used for automated tooth detection and numbering. 4545 pediatric panoramic X-ray images, processed in labelImg, were trained and tested in the Yolo algorithm. Results and Conclusions: The model was successful in detecting and numbering both primary and permanent teeth on pediatric panoramic radiographs with the mean average precision (mAP) value of 92.22 %, mean average recall (mAR) value of 94.44% and weighted-F1 score of 0.91. The proposed CNN method yielded high and fast performance for automated tooth detection and numbering on pediatric panoramic radiographs. Automatic tooth detection could help dental practitioners to save time and also use it as a pre-processing tool for detection of dental pathologies.

Artificial Intelligence Its Uses and Application in Pediatric Dentistry: A Review

“…AI models developed for application in pedodontics have mainly focused on: dental plaque on primary teeth ( n = 1) [ 21 ], ECC ( n = 6) [ 25 , 26 , 27 , 28 , 29 , 30 ], fissure sealant categorization ( n = 1) [ 31 ], mesiodens and supernumerary tooth identification ( n = 6) [ 3 , 10 , 22 , 23 , 24 , 41 ], chronological age assessment ( n = 4) [ 32 , 33 , 37 , 38 ], identification of deciduous and young permanent teeth ( n = 3) [ 34 , 35 , 39 ], children’s oral Health ( n = 2) [ 7 , 20 ], and ectopic eruption ( n = 2) [ 39 , 40 ] ( Figure 3 ).…”

“…Kaya, E. et al [ 39 ] tested the effectiveness of a deep learning system for automated tooth recognition and counting using YOLOv4, a CNN-based object identification model. The model was able to recognize and count both primary and permanent teeth.…”

“…Therefore, YOLO is an example of a single-stage detector utilized for fast and accurate object identification and categorization. YOLO stands out from other CNN algorithms because of its ability to do real-time object recognition while also achieving above-average results over a wide range of object classes on average [ 39 ].…”

“…Articles without complete texts, narrative reviews, scoping reviews, letters to the editor, opinion letters, case reports, brief communications, conference proceedings, and non-English language articles (84 articles) were all omitted ( Figure 1 ). Finally, only 25 papers met the qualifying requirements [ 3 , 7 , 10 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 ].…”

A Novel Collaborative Learning Model for Teeth and Fillings in Radiographs

“…Our proposed model is distinct from existing models in that it executes multiple tasks and generates a summary of each individual model’s findings. There have only been a few studies that have segmented primary teeth, and most of these have focused on identifying specific abnormalities (e.g., mesiodens) (Ahn et al 2021, Ha et al 2021, Jeon et al 2022, Kaya et al 2022, Kilic et al 2021, Pinheiro et al 2021). Moreover, while various deep learning methods have been developed to detect fillings, to our knowledge, the present model is the first to detect fillings in both primary and permanent teeth.…”