Two-step pipeline for oral diseases detection and classification: a deep learning approach
This proof-of-concept study demonstrates the feasibility of a two-step artificial intelligence (AI) pipeline combining object detection and classification to support early diagnosis of oral diseases. However, caution is warranted when interpreting the results of two-step approaches, as images missed by YOLO during detection are excluded from the classification stage, which may affect the reported performance metrics.
Authors
- Anthony Gonçalvès (ORCID: https://orcid.org/0000-0001-7570-7439)
- Márcio Ajudarte Lopes (ORCID: https://orcid.org/0000-0001-6677-0065)
- Cristina Saldivia‐Siracusa (ORCID: https://orcid.org/0000-0002-2422-2164)
- Alan Roger Santos‐Silva (ORCID: https://orcid.org/0000-0003-2040-6617)
- Anna Luíza Damaceno Araújo (ORCID: https://orcid.org/0000-0002-3725-8051)
- Marcos G. Quiles (ORCID: https://orcid.org/0000-0001-8147-554X)
- Arnaldo Silva
- Pablo Vargas (ORCID: https://orcid.org/0000-0002-6475-494X)
- Daniel Lobato Ferreira Ferraz (ORCID: https://orcid.org/0000-0003-3740-5127)
- André C. P. L. F. de Carvalho (ORCID: https://orcid.org/0000-0002-4765-6459)
- Paulo Rogério Ferreti Bonan (ORCID: https://orcid.org/0000-0002-4449-4343)
- Ivan José Correia Neto (ORCID: https://orcid.org/0009-0007-3828-0107)
- Camila Barcellos Calderipe (ORCID: https://orcid.org/0000-0003-0188-7435)
Institutions
- Universidade Federal da Paraíba (BR)
- Universidade de São Paulo (BR)
- Universidade Estadual de Campinas (UNICAMP) (BR)
- AC Camargo Hospital (BR)
- Methodist University of Piracicaba (BR)
- Universidade Federal de São Paulo (BR)
Publication Details
- Journal
- Frontiers in Oral Health
- Published
- 2025-10-27
- DOI
- https://doi.org/10.3389/froh.2025.1659323
- Citations
- 1
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 2.48
Funders
- Fundação de Amparo à Pesquisa do Estado de São Paulo