Pediatric Dental Disease Detection Using X-Ray Image Enhancements and Deep Learning Algorithms

The identification of pediatric dental diseases from panoramic radiographs is a challenging task due to the anatomical and positional differences of children’s oral structures. The present study attempts to solve this issue using a novel panoramic radiograph dataset for pediatric dental diseases. This is important because improving the standard of healthcare provided to pediatric patients and the prevention of dental complications later in life depend on the early identification of the problem and the active steps being taken to prevent it. The proposed approach employs a more systematic workflow that starts with image enhancement methods so that loss of image quality is minimized and salient characteristics are accentuated. These include the use of Contrast- Limited Adaptive Histogram Equalization (CLAHE), Unsharp Masking, Gamma Correction, and Min-Max Normalization. Next, for detection and classification purposes, the deep learning algorithms YOLOv11 and Faster R-CNN are employed in the image processing tasks of object localization and classification, respectively, owing to their unique advantages toward a reliable and accurate diagnosis of dental anomalies. Of the models trained in the pediatric radiograph data set edited adequately with the marks that pinpoint the relevant details, the clearly annotated data set allowed YOLOv11 to outperform its rivals, recording a precision of 0.965 while achieving a recall of 0.92 and a mean Average Precision score of 0.976. Although Faster R-CNN recorded a higher recall, the precision of YOLOv11 was more impressive; therefore, it was more consistent in providing results in the identification of early stages of the diseases. This method helps to fill the gaps in the existing literature on the interpretation of pediatric radiographs, and it helps to offer a scalable and noninvasive diagnostic method which can help to improve the efficiency of clinical procedures.