Estimating the bone quality of jaw in panoramic images for dental implant placement using artificial intelligence

Abstract

Introduction: One of the crucial parameters affecting the success of dental implant treatments is the quality of the bone. Therefore, estimation of the bone quality before placement of the dental implant can affect prognosis and treatment. The aim of the study was to estimate the quality of jaw bone in panoramic images for implant placement using artificial intelligence. Materials and Methods: Panoramic and CBCT radiographs of jaws will be gathered from the archive of Dentistry Faculty of Tabriz University of Medical Sciences. Bone quality will be separated and labeled based on Misch’s classification. In the next step, the labeled panoramic radiographs will be provided to the relevant expert to build an artificial intelligence model. The total sample size required for this model is at least 352 images. In this method, panoramic images of the jaws will be defined as input for the model. After conducting the confusion matrix, the validity of the designed model will be assessed by diagnostic parameters which includes sensitivity, specificity, accuracy, positive predictive value, negative predictive value. Final results will be reported as mentioned diagnostic parameters. Results: The trained netmork in this study estimated bone quality with an error of 0.7 and an accuracy of 75%. The network evaluation was 11% for class D1, 72% for class D2, 28% for class D3 and 77% for class D4. The designed network showed the highest accuracy and recognition for class D4 (0.63 and 0.77, respectively) and the lowest accuracy and recognition for class D1 (0.5 and 0.11, respectively). Conclusion: The artificial intelligence model trained in this study can receive cropped panoramic images of edentulous areas and classify it into one of four groups according to the Misch’s classification. This model has more diagnostic accuracy to determine the D4 bone quality.