Deep Learning on Histology Images for Differentiating Fibro-osseous Lesions of the Jaw - PubMed
5 hours ago
- #deep learning
- #fibro-osseous lesions
- #computational pathology
- Deep learning model developed to differentiate fibro-osseous lesions of the jaw (FD, COF, COD) using histology images.
- Model trained on 1,218 whole slide images from 338 cases across 3 institutions.
- Weakly supervised multislide model achieved best performance (AUC: 0.86, accuracy: 0.71), surpassing oral pathologists' diagnostic accuracy.
- Model identified key histomorphologic patterns via probability heat maps for lesion differentiation.
- Potential as a supportive tool alongside clinical, radiologic, and molecular data highlighted.
- Limitations include small test cohort size and lack of geographic diversity, necessitating further validation.