EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

RiTex: Harmonization of Radiomic Features Based on Riemannian Geometry

Darya A Voitenko, Anton V Vladzymyrskyy iD, Olga V Omelyanskaya iD, Yuriy A Vasilev iD, Ivan A Blokhin iD, Maria R Kodenko iD

Journal of Imaging · 17.06.2026 · 12 (6) · 264

DOI: 10.3390/jimaging12060264 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Imaging, online erschienen am 17.06.2026. Verfasst von Darya A Voitenko und 5 weiteren Autor:innen. Beteiligte Einrichtung: State Budget-Funded Health Care Institution of the City of Moscow "Research and Practical Clinical Center for Diagnostics and Telemedicine Technologi…. Die Arbeit nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Batch effects arising from variations in hardware, acquisition protocols, and reconstruction parameters present a critical challenge in radiomics, limiting the generalizability of models across multicentre studies. Existing harmonization methods, such as ComBat, CovBat, z-score normalization, and Generative Adversarial Networks, exhibit significant limitations when applied to high-dimensional radiomic data. ComBat assumes a linear feature space and tends to leave residual center-specific information recoverable by downstream classifiers. This paper introduces RiTex (Riemannian Texture Harmonization), a framework that solves a generalized eigenvalue problem between class-aware biological scatter and Ledoit–Wolf-regularized per-batch covariances, with the SPD-manifold Fréchet mean used as a principled averaging step. We evaluate RiTex on the 50-dataset radMLBench benchmark and on a new four-center head-and-neck benchmark with known center labels (n = 380 patients, k = 4 centers from TCIA: HGJ, MDACC, Maastro, QIN). On radMLBench, RiTex reduces the batch auto-detection AUC in 48/50 (96%) datasets, 42/50 (84%) reductions remain significant after Benjamini–Hochberg correction; the mean Batch AUC reduction is ΔBatch = −0.365 (95% bootstrap CI [−0.418, −0.312]), with no significant degradation in biological AUC (mean ΔBio = +0.018, 95% CI [−0.011, +0.047]). On the H&N benchmark with real center labels, RiTex reduces the Batch AUC from 0.74 to 0.59, while ComBat and CovBat leave it at ≈0.98. A component-wise ablation shows that the dominant source of empirical performance is the GEVD step, together with Ledoit–Wolf shrinkage. The SPD Fréchet mean acts as a theoretical scaffold with a negligible empirical contribution (ΔBatch AUC = −0.014 vs. arithmetic mean).

Schlagwörter

Stichwörter der Autor:innen

ComBatCovBatFréchet meanRiemannian geometrySPD matricesbatch effectsgeneralized eigenvalue problemharmonizationmulti-center validationradiomics

Autor:innen und Institutionen

  • State Budget-Funded Health Care Institution of the City of Moscow "Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of the Moscow Health Care Department", 127051 Moscow, Russia.
  • Federal State-Funded Educational Institution of Higher Education, Lomonosov Moscow State University, 119991 Moscow, Russia.

ORCID-Profile: Anton V Vladzymyrskyy, Olga V Omelyanskaya, Yuriy A Vasilev, Ivan A Blokhin, Maria R Kodenko

Zitieren

APA

Voitenko, D. A., Vladzymyrskyy, A. V., Omelyanskaya, O. V., Vasilev, Y. A., Blokhin, I. A., & Kodenko, M. R. (2026). RiTex: Harmonization of Radiomic Features Based on Riemannian Geometry. Journal of Imaging, 12(6), 264. https://doi.org/10.3390/jimaging12060264

Vancouver

Voitenko DA, Vladzymyrskyy AV, Omelyanskaya OV, Vasilev YA, Blokhin IA, Kodenko MR. RiTex: Harmonization of Radiomic Features Based on Riemannian Geometry. Journal of Imaging. 2026;12(6):264. doi:10.3390/jimaging12060264

BibTeX
@article{Voitenko2026,
  author = {Voitenko, Darya A and Vladzymyrskyy, Anton V and Omelyanskaya, Olga V and Vasilev, Yuriy A and Blokhin, Ivan A and Kodenko, Maria R},
  title = {RiTex: Harmonization of Radiomic Features Based on Riemannian Geometry},
  journal = {Journal of Imaging},
  year = {2026},
  volume = {12},
  number = {6},
  pages = {264},
  doi = {10.3390/jimaging12060264}
}
Kostenlos starten →