Deep Learning-Based Diagnosis of Temporomandibular Joint Osteoarthritis Using Whole-Body Bone Scans
Published in iScience, 2025
Authors: Yeon-Hee Lee, Hee-Sung Kim, Seonggwang Jeon, Q-Schick Auh, Il Ki Hong, Sunju Choi, Fernando Guastaldi, Hyungsoon Im, Yung-Kyun Noh, Akhilanand Chaurasia
Venue: iScience, 2025
This study developed deep learning models based on the VGG16 architecture to automatically diagnose temporomandibular joint osteoarthritis (TMJ-OA) from bone scintigraphy. Using a dataset of 1,943 patients (3,886 TMJs), the VGG16-Lite model achieved diagnostic accuracy of AUC > 0.90 on head-and-neck imaging across age and sex subgroups, outperforming pretrained models.
My contributions: Manuscript writing, data analysis, data interpretation, and figure preparation.