Detection of Dens Invaginatus on Panoramic Radiographs Using Deep Learning Algorithms

Abstract

BackgroundDens invaginatus is a developmental dental anomaly characterized by enamel folding into the dental papilla during odontogenesis. Early detection allows for appropriate management and reduces the risk of complex treatments.AimThis study aimed to evaluate the success and reliability of YOLOv5 and YOLOv8 deep learning models with two different labeling methods for the detection of teeth with dens invaginatus in panoramic radiographs.DesignIn this study, 656 panoramic radiographs of patients aged 8 to 18 were labeled for teeth with dens invaginatus in the anterior region using segmentation and detection methods. The labeling of the images was performed using the CranioCatch software (Eski & scedil;ehir, Turkey) by two pediatric dentists.ResultsThe YOLOv5 model achieved a precision, sensitivity, and F1-score of 0.945, 0.887, and 0.915 for detection, and 0.905, 0.928, and 0.916 for segmentation, respectively. The precision, sensitivity, and F1-score of the YOLOv8 model with the detection method were 0.950, 1, and 0.974, respectively, while these values were 0.940, 0.994, and 0.966 for the segmentation method.ConclusionIt was observed that the YOLO models were successful in detecting dens invaginatus. It is believed that deep learning-supported systems could be integrated into pediatric dentistry practice and serve as a decision support mechanism.