Accuracy of Artificial Intelligence-Assisted Staging and Grading for Diagnosis of Periodontitis: A Cross-Sectional Study
Accuracy of Artificial Intelligence-Assisted Staging and Grading for Diagnosis of Periodontitis: A Cross-Sectional Study
This observational study aims to develop and assess the accuracy, specificity, and sensitivity of a deep learning model for the classification of periodontitis using panoramic radiographs and clinical data inputs. A total of 341 panoramic images will be retrospectively collected and labeled by experienced periodontists to train and test the model. The model will be evaluated for its ability to determine the stage and grade of periodontitis based on the 2017 classification guidelines set by the American Academy of Periodontology. The results will be compared to those of clinical experts to validate the AI-assisted diagnostic system. This study is conducted at the Faculty of Dentistry, Ain Shams University, in fulfillment of a Master's degree in Periodontology.
a convolutional neural network (CNN)-based deep learning model will be trained using 341 panoramic radiographs and relevant clinical data to classify patients according to stage and grade of periodontitis. Images will be obtained from the Oral and Maxillofacial Radiology Department at Ain Shams University. The inclusion criteria includes radiographs of patients with periodontal bone loss and radiographs of patients with orthodontic brackets, mixed dentition, and artifacts will be excluded. Clinical data, including age, diabetes status, and smoking history, will be incorporated to calculate grading using the bone loss/age ratio of the testing set.
The collected dataset will be divided into 80% for training and 20% for testing. Six anatomical landmarks will be annotated per tooth to calculate the percentage of bone loss mesially and distally, which will be used to determine the stage of disease. Grading will be determined based on percentage bone loss relative to patient age and systemic modifiers. Expert-labeled datasets will serve as a reference standard for evaluating the performance of the AI model.
The primary objective is to evaluate the model's diagnostic accuracy for staging and grading compared to specialist assessments. The secondary objective is to measure the sensitivity and specificity of the model.
Inclusion Criteria:
Exclusion Criteria:
x-ray images with