Supervised Learning for Accurate Tooth Segmentation and Numbering in Panoramic X-Ray Images

This article aims to develop an innovative method for segmenting and numbering teeth in panoramic dental X-ray images, which are crucial for dental diagnosis and treatment planning. However, segmenting and numbering teeth in these images is challenging due to noise, artifacts, and variations in tooth shape and size. To tackle this issue, the project utilizes a combination of image processing and machine learning techniques to accurately segment and number teeth based on the standard dental notation system. The proposed method involves loading images into CVAT for manual tooth labeling, followed by extracting bounding box coordinates to overlay them on test images. The method's performance is evaluated using a dataset of X-ray images, achieving an accuracy of 86.7% and a sensitivity of 84.4%. While the results demonstrate the effectiveness of this supervised approach, further improvements are needed in automatic tooth detection and segmentation precision. Future research will focus on enhancing segmentation quality and automating the annotation of individual tooth segments.