Towards Universal Cytopathology Segmentation: Transformer and Cnn Models Across Stains
The increasing demand for accurate and efficient cytopathological diagnostics at early stages highlighted the need for computational solutions of handling real-world clinical variability. Our study evaluates state-of-the-art convolutional and transformer-based models on three distinct cytological stains: <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{Pa}{-}$</tex> panicolaou, Feulgen and AgNOR. By comparing models trained on single-stain datasets with those trained on combined multistain dataset, we demonstrate that it improves generalization and operational efficiency.
Authors
- Aldo von Wangenheim (ORCID: https://orcid.org/0000-0003-4532-1417)
- Fabiana Botelho de Miranda Onofre (ORCID: https://orcid.org/0000-0003-4857-5770)
- Alexandre Sherlley Casimiro Onofre (ORCID: https://orcid.org/0000-0003-1883-6087)
- Luís Otávio Santos (ORCID: https://orcid.org/0000-0002-6660-462X)
- Douglas Dyllon Jerônimo de Macedo (ORCID: https://orcid.org/0000-0002-3237-4168)
- Rodrigo P. S. Ribeiro (ORCID: https://orcid.org/0000-0002-6112-8660)
- Bruno J. Lopes
- Bibiana Q. Tiellet
Institutions
- Universidade Federal de Santa Catarina (BR)
Publication Details
- Published
- 2025-06-18
- DOI
- https://doi.org/10.1109/cbms65348.2025.00077
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina