Deep Learning and Image Processing Techniques applied in Panoramic X-Ray Images for Teeth Detection and Dental Problem Classification

Abstract

Due to its remarkable achievements in detection,prediction, and classification, deep convolutional neural network has gotten huge attention into clinical analysis. Specialists can use panoramic dental radiograph analysis to uncover issues in low-light locations, inside the buccal cavities, or even narrow regions. Weak picture resolution or weariness, on the other hand, can lead the diagnosis to differ, making therapy more difficult. With the help of comprehensive x-ray pictures, the automated tooth detection and dental condition classification method we present in this study, medical practitioners can make more accurate diagnosis decisions. In order to conduct this study, panoramic radiographics from three dental offices were gathered and interpreted, showcasing 14 different potential tooth problems. An annotated dataset was used to train CNN and acquire semantic segmentation data. Then, different image processing methods were applied to segment and fine-tune the bounding boxes corresponding to the teeth defections. The final step involved labelling each tooth sample inside the identified area of interest and using histogram-based majority voting to determine the issues that affected it. Some of the criteria used to assess the developed method included specificity, memory, result for derived bounded area detection, and intersection over union in supervised classification. The outcomes were contrasted with information from different approaches and found to be superior, demonstrating the proposed solutions’ superiority. In addition, tasks such as tooth categorization, identification of illnesses and severe gum disease like periodontitis, and how much cavity cleaning should be done are completed.