EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Nested Cross-Validation Reveals Performance Inflation in MRI Radiomics for Early Mortality Prediction in IDH-Wildtype Glioblastoma

Lucas I Becker iD, Nicolas Noel Neidert, Roberto Doria-Medina iD, Manou Overstijns iD, Maryam Wendeberg, Urs Würtemberger iD, Horst Urbach iD

Cancers · 12.08.2026 · 18 (16) · 2595

DOI: 10.3390/cancers18162595 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Cancers, online erschienen am 12.08.2026. Verfasst von Lucas I Becker und 6 weiteren Autor:innen. Beteiligte Einrichtung: Department of Neuroradiology, University Medical Centre Freiburg, 79100 Freiburg, Germany. Die Arbeit nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Glioblastoma (GBM) carries a median survival of 15 months. Approximately 16–27% of patients die within six months despite standard therapy. Radiomic MRI features have been proposed as prognostic biomarkers, yet many published studies employ standard cross-validation (CV) with feature selection on the full dataset, introducing data leakage.

Methods

Sixty patients with IDH-wildtype GBM from the publicly available UCSF-PDGM dataset (single-center, GE Discovery MR750, 3 T; 2015–2021) with BraTS 2021 segmentation masks were analyzed. One thousand two hundred eighty-four (1284) IBSI-compliant radiomic features (7 feature classes × 4 MRI sequences × 3 tumor subregions) were extracted using PyRadiomics. After preprocessing (variance filter, correlation filter |r| > 0.95), approximately 120 features remained per fold. Feature selection was performed strictly within each training fold of a nested 5-fold cross-validation framework (5-fold × 3 repeats = 15 outer folds).

Results

For 1-year mortality, radiomics AUC dropped from 0.816 (standard CV) to 0.593 (nested CV; ΔAUC = −0.223, 27% inflation), while clinical models remained stable (0.708 vs. 0.705). For early mortality (≤180 days, n = 16 events), standard CV inflated radiomics AUC to 0.888, whereas nested CV yielded 0.815 (ΔAUC = −0.073, 8% inflation). In feature stability analysis, whole-tumor surface area (13/15 folds) and mesh volume (10/15 folds) showed the highest cross-endpoint stability; nine of 13 exploratory OS-associated features were never selected for early mortality classification. Extent of resection (HR = 0.44, p = 0.009) and age (HR = 1.03, p = 0.038) were independently associated with overall survival; tumor surface area remained independently associated with survival (HR = 1.38, p = 0.021).

Conclusions

Nested cross-validation revealed substantial performance inflation in standard radiomics pipelines. These descriptive inflation estimates are specific to this dataset and pipeline configuration and should not be generalized as universal parameters for radiomics.

Schlagwörter

Stichwörter der Autor:innen

MRIdata leakageearly mortalityglioblastomamachine learningnested cross-validationprognostic biomarkersradiomics

Autor:innen und Institutionen

  • Department of Neuroradiology, University Medical Centre Freiburg, 79100 Freiburg, Germany.
  • Department of Neurosurgery, University Medical Centre Freiburg, 79100 Freiburg, Germany.
  • Raya Diagnostics, 80802 Munich, Germany.

ORCID-Profile: Lucas I Becker, Roberto Doria-Medina, Manou Overstijns, Urs Würtemberger, Horst Urbach

Zitieren

APA

Becker, L. I., Neidert, N. N., Doria-Medina, R., Overstijns, M., Wendeberg, M., Würtemberger, U., & Urbach, H. (2026). Nested Cross-Validation Reveals Performance Inflation in MRI Radiomics for Early Mortality Prediction in IDH-Wildtype Glioblastoma. Cancers, 18(16), 2595. https://doi.org/10.3390/cancers18162595

Vancouver

Becker LI, Neidert NN, Doria-Medina R, Overstijns M, Wendeberg M, Würtemberger U, et al. Nested Cross-Validation Reveals Performance Inflation in MRI Radiomics for Early Mortality Prediction in IDH-Wildtype Glioblastoma. Cancers. 2026;18(16):2595. doi:10.3390/cancers18162595

BibTeX
@article{Becker2026,
  author = {Becker, Lucas I and Neidert, Nicolas Noel and Doria-Medina, Roberto and Overstijns, Manou and Wendeberg, Maryam and Würtemberger, Urs and Urbach, Horst},
  title = {Nested Cross-Validation Reveals Performance Inflation in MRI Radiomics for Early Mortality Prediction in IDH-Wildtype Glioblastoma},
  journal = {Cancers},
  year = {2026},
  volume = {18},
  number = {16},
  pages = {2595},
  doi = {10.3390/cancers18162595}
}
Kostenlos starten →