Evaluating Artificial Intelligence Accuracy in Gender Prediction Using Mandible Measurements from 3D CBCT Scans of Adult Egyptian Patients

What this Trial Is About

Researchers are evaluating the accuracy of artificial intelligence AI in identifying gender using linear morphometric measurements of the mandible from Cone Beam Computed Tomography CBCT scans of adult Egyptian dental patients. This observational study aims to validate AI as a reliable and objective tool for gender determination, which is important in forensic odontology. The study compares AI predictions to known patient gender records and assesses the precision of AI in anatomical point identification. The study uses pre-existing CBCT scans, where the mandible is automatically segmented using imaging software. A radiologist manually measures specific anatomical points, and these measurements serve as the reference standard. A three-dimensional deep learning model 3D U-Net is trained to locate anatomical points and perform measurements automatically. Then, artificial neural networks and logistic regression models predict gender based on these measurements. The dataset is divided for training, validation, and testing, with cross-validation if needed. Participants are not actively involved as the study uses anonymized scans. The radiologist performs manual measurements blinded to patient gender. The AI models performance is evaluated by comparing its anatomical point placements and gender predictions against the radiologists measurements and known gender, using metrics such as accuracy, precision, recall, and error distances. The study duration includes data analysis until the primary completion date, with no missing data expected due to its retrospective nature.

Validating AI for Gender Prediction Using Morphometric Analysis of the Mandible