Dental Disease Detection Based On Deep Learning Algorithm Using Various Radiographs

Abstract

Dental Diseases Such As Dental Caries, Periodontal Disease, Periapical Lesions, And Bone Loss Are Among The Most Prevalent Oral Health Issues Globally. Early And Accurate Detection Is Critical To Preventing Progression And Ensuring Effective Treatment. Traditional Diagnostic Methods Relying On Manual Inspection Of Radiographs Are Often Time-consuming And Subject To Variability In Interpretation. This Study Presents A Deep Learning-based Approach For The Automated Detection And Classification Of Dental Diseases Using Various Types Of Radiographic Images, Including Panoramic, Periapical, And Bitewing Radiographs. A Custom Convolutional Neural Network (CNN) Architecture, As Well As Pre-trained Models Such As ResNet50 And EfficientNet, Were Trained And Evaluated On A Curated Dataset Comprising Over 5,000 Labeled Radiographs Annotated By Experienced Dental Professionals. Image Preprocessing Techniques, Such As Contrast Enhancement And Noise Reduction, Were Applied To Improve Image Quality And Model Performance. The Models Were Trained To Classify Multiple Disease Conditions, Including Caries, Periodontal Bone Loss, Impacted Teeth, Cysts, And Periapical Abscesses. The Proposed Models Demonstrated High Classification Accuracy, With The Best-performing Model Achieving An Overall Accuracy Of 93.2%, Sensitivity Of 91.5%, And Specificity Of 94.8% Across All Radiograph Types. Cross-validation Confirmed The Model’s Robustness And Generalization Across Different Imaging Conditions And Patient Demographics. Furthermore, Class Activation Mapping (CAM) Was Used To Provide Visual Explanations, Increasing The Interpretability Of The Results And Enhancing Clinical Trust. This Study Confirms The Viability Of Deep Learning Systems As A Reliable Tool For Dental Disease Diagnosis. By Leveraging Multiple Radiographic Modalities, The System Enhances Diagnostic Accuracy And Can Serve As A Valuable Adjunct In Clinical Workflows, Reducing Diagnostic Delays And Improving Patient Outcomes.

Keywords