Multicollinearity and redundancy of the PET radiomic feature set
Question Radiomic feature sets contain many strongly correlating features, which results in statistical challenges. Findings Analysis of the correlation matrices showed that the same radiomic features were strongly correlated in five independent [<sup>18</sup>F]FDG-PET cohorts with different tumour types. Clinical relevance At least two-thirds of the radiomic feature set could be omitted, because of strong multicollinearity. More redundant features could be identified using a less conservative threshold.
Authors
- Wim J.G. Oyen (ORCID: https://orcid.org/0000-0001-8235-7078)
- Nicolas Aide (ORCID: https://orcid.org/0000-0001-9207-0847)
- Wyanne A. Noortman (ORCID: https://orcid.org/0000-0003-0450-951X)
- Renaud Lhommel (ORCID: https://orcid.org/0000-0003-0668-6122)
- Johan Bussink (ORCID: https://orcid.org/0000-0002-5751-4796)
- Erik H.J.G. Aarntzen (ORCID: https://orcid.org/0000-0002-3809-0182)
- Christophe M. Deroose (ORCID: https://orcid.org/0000-0002-6080-1577)
- Lioe‐Fee de Geus‐Oei (ORCID: https://orcid.org/0000-0003-1817-2743)
- Dennis Vriens (ORCID: https://orcid.org/0000-0001-5762-6766)
- Elizabeth J. de Koster (ORCID: https://orcid.org/0000-0001-8111-1172)
- Christophe Le Tourneau (ORCID: https://orcid.org/0000-0001-9772-4686)
- Erik Vegt (ORCID: https://orcid.org/0000-0001-7456-2258)
- Jelle P. H. Ruurda (ORCID: https://orcid.org/0000-0001-6584-1677)
- Floris H. P. van Velden (ORCID: https://orcid.org/0000-0003-2859-2119)
- Lianne Triemstra (ORCID: https://orcid.org/0000-0002-7057-0274)
- Tineke W.H. Meijer (ORCID: https://orcid.org/0000-0002-8550-6523)
Institutions
- Cliniques Universitaires Saint-Luc (BE)
- University Medical Center Groningen (NL)
- Humanitas University (IT)
- Radboud University Nijmegen (NL)
- Inserm (FR)
- Université Paris-Saclay (FR)
- Leiden University Medical Center (NL)
- Universitair Ziekenhuis Leuven (BE)
- Erasmus MC (NL)
- Radboud University Medical Center (NL)
- University Medical Center Utrecht (NL)
- Centre François Baclesse (LU)
- Centre François Baclesse (FR)
- Rijnstate Hospital (NL)
- Institut Curie (FR)
- University of Tübingen (DE)
- University of Twente (NL)
Publication Details
- Journal
- European Radiology
- Published
- 2025-05-07
- DOI
- https://doi.org/10.1007/s00330-025-11637-7
- Citations
- 4
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 5.44
Funders
- Université Catholique de Louvain
- Institut National de la Santé et de la Recherche Médicale
- Radboud Universiteit
- Universitair Medisch Centrum Groningen
- Radboud Universitair Medisch Centrum
- Universitaire Ziekenhuizen Leuven, KU Leuven