Interobserver ground-truth variability limits performance of automated glioblastoma segmentation on [18F]FET PET
Incorporating a threshold map can significantly improve tumor segmentation performance when it aligns well with the ground-truth label. However, the current inability to reliably reproduce these maps-both manually and automatically-or the ground-truth tumor labels, restricts the achievable accuracy for automated glioblastoma segmentation on [<sup>18</sup>F]FET PET, highlighting the need for more consistent definitions of such ground-truth delineations.
Authors
- Hendrik Everaert (ORCID: https://orcid.org/0000-0003-0756-0089)
- Sophie Bourgeois
- Ine Dirks (ORCID: https://orcid.org/0000-0002-1648-0358)
- Johnny Duerinck (ORCID: https://orcid.org/0000-0001-9869-9806)
- S. De Sutter (ORCID: https://orcid.org/0000-0003-1236-1420)
- Wietse Geens (ORCID: https://orcid.org/0000-0002-2496-9209)
- Jef Vandemeulebroucke (ORCID: https://orcid.org/0000-0001-5714-3254)
- Laurens Raes (ORCID: https://orcid.org/0000-0003-4544-7001)
Institutions
- Vrije Universiteit Brussel (BE)
- IMEC (BE)
- Universitair Ziekenhuis Brussel (BE)
Publication Details
- Journal
- EJNMMI Physics
- Published
- 2025-06-06
- DOI
- https://doi.org/10.1186/s40658-025-00767-y
- Citations
- 4
- Primary Topic
- Medical Imaging Techniques and Applications
- Type
- article
- Field-Weighted Citation Impact
- 5.44
Funders
- Vlaams Supercomputer Centrum
- Fonds Wetenschappelijk Onderzoek
- Vlaamse regering
- Horizon 2020