EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Validation Is Not Enough: A Duplicate-Aware, Reproducible Frozen-Feature Benchmark for Brain Tumour MRI Classification on BRISC 2025

Narasimha Rao Vajjhala, Ervin Ramollari, Rajesh Kumar

Information · 19.11.2026 · 17 (11) · 1175

DOI: 10.3390/info17111175 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Information, online erschienen am 19.11.2026. Verfasst von Narasimha Rao Vajjhala und 2 weiteren Autor:innen. Die Arbeit nennt 78 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Public brain tumour MRI collections are routinely reported with near-perfect classification accuracy, yet they carry no patient identifiers and their train/test overlap is rarely quantified. We present a duplicate-aware, reproducible frozen-feature benchmark on BRISC 2025 (6000 contrast-enhanced T1-weighted slices; glioma, meningioma, pituitary tumour, no tumour). A perceptual-hash audit found 1352 candidate near-duplicate pairs, with 439 of them crossing the official train/test split, and 104 training images that are pixel-identical to test images; a visual review of sampled pairs showed that most candidates were other slices of the same scan, with an estimated 5–10% false matches. Removing redundant copies and every training image connected to the test set left 5477 images, a split that is duplicate-aware but not patient-independent. Under an identical linear probe, ImageNet-supervised ResNet-18/50 and self-supervised DINOv2 ViT-S/14 were competitive: DINOv2 was numerically ahead (macro-F1 77.3 ± 0.1%, macro-AUROC 93.2%), but no pairwise difference survived Holm correction. In-pool validation overstated official test macro-F1 by 9.7–13.5 points; control experiments showed that this gap is not an artefact of the filtering but mainly reflects a distribution difference between the official partitions. Meningioma recall stayed at 50–53%, and all linear heads were under-confident. We release the splits, audit tables and per-image predictions.

Zitieren

APA

Vajjhala, N. R., Ramollari, E., & Kumar, R. (2026). Validation Is Not Enough: A Duplicate-Aware, Reproducible Frozen-Feature Benchmark for Brain Tumour MRI Classification on BRISC 2025. Information, 17(11), 1175. https://doi.org/10.3390/info17111175

Vancouver

Vajjhala NR, Ramollari E, Kumar R. Validation Is Not Enough: A Duplicate-Aware, Reproducible Frozen-Feature Benchmark for Brain Tumour MRI Classification on BRISC 2025. Information. 2026;17(11):1175. doi:10.3390/info17111175

BibTeX
@article{Vajjhala2026,
  author = {Vajjhala, Narasimha Rao and Ramollari, Ervin and Kumar, Rajesh},
  title = {Validation Is Not Enough: A Duplicate-Aware, Reproducible Frozen-Feature Benchmark for Brain Tumour MRI Classification on BRISC 2025},
  journal = {Information},
  year = {2026},
  volume = {17},
  number = {11},
  pages = {1175},
  doi = {10.3390/info17111175}
}
Kostenlos starten →