Impact of annotation imperfections and auto-curation for deep learning-based organ-at-risk segmentation
Up to 30% case removal, the curation benefit outweighed the training variance lost through curation. Considering the notable ease of implementation, high sensitivity in simulations and performance gains already at lower curation fractions, as a conservative middle ground, we recommend 15% curation of training cases when training CNNs using clinical PG contours.
Authors
- Oliver J. Gurney‐Champion (ORCID: https://orcid.org/0000-0003-1750-6617)
- Ben J. Slotman (ORCID: https://orcid.org/0000-0002-9902-4592)
- Victor I.J. Strijbis
- W. Verbakel
Institutions
- Dutch Cancer Society (NL)
- Siemens Healthcare (United States) (US)
- Cancer Center Amsterdam (NL)
- Varian Medical Systems (United States) (US)
Publication Details
- Journal
- Physics and Imaging in Radiation Oncology
- Published
- 2024-10-01
- DOI
- https://doi.org/10.1016/j.phro.2024.100684
- Citations
- 4
- Primary Topic
- Machine Learning and Data Classification
- Type
- article
- Field-Weighted Citation Impact
- 1.39
Funders
- Varian Medical Systems
- Nederlandse Organisatie voor Wetenschappelijk Onderzoek
- KWF Kankerbestrijding