Metrics comparison between PyTorch and TensorRT models
| Model | PyTorch results | TensorRT results | ||
|---|---|---|---|---|
| Dice | IoU | Dice | IoU | |
| SegFormer-B2 | 0.9787 | 0.9588 | 0.9231 | 0.8684 |
| SegFormer-B4 | 0.9843 | 0.9694 | 0.9433 | 0.9025 |
| SSFormer-Small | 0.1659 | 0.1590 | 0.1606 | 0.1449 |
| SSFormer-Large | 0.1780 | 0.1616 | 0.1667 | 0.1487 |
IoU: Intersection over Union
We extend our gratitude to all partners and collaborators who contributed to the successful implementation and validation of the proposed system, including Key To Business, Department of Neurosciences – Catholic University of the Sacred Heart, and Studio5T.
MM: Conceptualization, Methodology, Data curation, Investigation, Formal analysis, Writing—original draft. SS: Software, Investigation, Formal analysis. MP: Formal analysis, Writing—review & editing. IGC: Project administration, Formal analysis, Supervision, Conceptualization. All authors read and approved the final version of the manuscript.
The authors declare that they have no conflicts of interest.
The datasets used in this manuscript were sourced from Kvasir-SEG, CVC-ClinicDB, ETIS-LaribPolypDB, PolypGen, CVC-ColonDB, CVC-300, and PolypDataset-TCNoEndo. All datasets and information presented in this article are fully anonymized and do not contain any personally identifiable information. Therefore, ethical approval, consent to participate, and consent to publication are not required.
Not required.
Not required.
The dataset used in this study is derived from publicly available sources cited in the manuscript. The details of each dataset are as follows: (1) Kvasir-SEG: Simula Datasets - Kvasir SEG; (2) CVC-ClinicDB: Simula Datasets - Kvasir SEG; (3) ETIS-LaribPolypDB: ETIS-LaribPolypDB; (4) PolypGen: Simula Datasets - Kvasir SEG; (5) CVC-ColonDB: CVC colon DB | Visual Interaction Group; (6) CVC-300: CVC-300; (7) PolypDataset-TCNoEndo: An augmented version of Kvasir-SEG. These datasets were selected to ensure a diverse and representative collection of polyp appearances, enhancing model generalization and robustness for real-world clinical applications. However, due to current project constraints, the integrated dataset cannot be openly released at this time. Access may be granted upon reasonable request and will be evaluated on a case-by-case basis.
This project has been co-financed by the European Union through the PR FESR 2021–2027 RSI program of Regione Lazio, managed by LazioInnova [CUP F89J23001090007], and approved with the publication of the rankings related to the public notice “Riposizionamento competitivo RSI PR FESR 2021–2027 Regione Lazio” in the BUR on 21/11/2023. The authors would like to thank the European Union and Regione Lazio for their support in enabling this research. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
© The Author(s) 2025.
Open Exploration maintains a neutral stance on jurisdictional claims in published institutional affiliations and maps. All opinions expressed in this article are the personal views of the author(s) and do not represent the stance of the editorial team or the publisher.