Multi-site Radiomics and Machine Learning for Bone Status Classification in Panoramic Radiographs: A STARD-Compliant Diagnostic Accuracy Study
This STARD-compliant study demonstrates that texture analysis of panoramic radiographs, particularly when processed with Random Forest models, holds potential as an opportunistic screening tool for detecting early postmenopausal bone loss in women, despite limitations related to small sample size and variable model performance across different bone status classes.