Diagnosing oral and maxillofacial diseases using deep learning
Source check: Source checked. Not independent scientific confirmation.
Finding: DOLNet, a multi-scale hierarchical-attention network, classifies and localizes odontogenic lesions on panoramic radiographs. The paper reports recall and F1 gains of up to 42.4% and 44.2% over recent methods, and 10.7% and 10.8% higher recall and F1 than clinicians. The authors note limited dataset access and generalization.
Technology: DOLNet hierarchical-attention neural network
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- DOLNet uses mutually influencing hierarchical attention across image scales to classify and localize odontogenic lesions in panoramic radiographs; the paper compares performance with clinicians and other methods.
- Test and result
- DOLNet was evaluated on panoramic radiographs for odontogenic-lesion classification and localization and compared with recent methods and human clinicians. The paper reports recall and F1 gains of up to 42.4% and 44.2% over recent methods, and 10.7% and 10.8% higher recall and F1 than clinicians.
- Limitation / still unknown
- The authors identify limited access to large, well-annotated datasets; performance beyond the evaluated images and clinical settings is not established.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://europepmc.org/article/MED/38291068 · https://www.nature.com/articles/s41598-024-52929-0 · https://pubmed.ncbi.nlm.nih.gov/38291068/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.