Harmonization Strategies for Handcrafted Radiomic Features: A Narrative Review of Reproducibility Challenges
Saudi Journal of Health Research & Practice · 06.05.2026 · 2 (3 & 4) · 26-34
DOI: 10.63908/ekk5ja44 ↗Kurz gefasst
Fachartikel in der Zeitschrift Saudi Journal of Health Research & Practice, online erschienen am 06.05.2026. Verfasst von Majed H. Salawi und 10 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Radiomics has emerged as a powerful approach for extracting quantitative features from medical images to support clinical decision-making in precision medicine. However, the reproducibility and generalizability of radiomic features (RFs) remain significant challenges, as these features are highly sensitive to variations in image acquisition protocols, reconstruction parameters, scanner types, and processing methodologies across different institutions. This variability limits the clinical translation of radiomics-based models and their applicability in multicentre studies. This review aims to summarize current harmonization strategies for handcrafted radiomic features and evaluate their role in improving reproducibility across multicentre imaging studies. Despite the growing number of proposed harmonization techniques, no consensus currently exists regarding optimal strategies for improving radiomic feature reproducibility across heterogeneous imaging datasets. Harmonization techniques have been developed to address these limitations and improve the consistency of RFs across heterogeneous imaging datasets. This narrative review synthesizes evidence from published radiomics studies focusing on both image-level and feature-level harmonization approaches. Image-level harmonization approaches include standardizing acquisition protocols and applying preprocessing techniques such as resampling and filtering, while feature-level harmonization encompasses statistical methods including z-score normalization, histogram matching, and ComBat harmonization. Evidence from multiple studies demonstrates that ComBat harmonization can effectively reduce inter-scanner variability and improve model performance in multicentre settings. The review also addresses quality assessment frameworks, including the Radiomics Quality Score and Image Biomarker Standardization Initiative guidelines. While significant progress has been made in developing harmonization methods, further research is needed to establish radiomics-specific approaches that account for imaging parameter differences and the inherent nature of different radiomic features, as this would enable the robust clinical implementation of radiomics in routine practice.
Zitieren
APA
Salawi, M. H., Khawaji, K. I., Mohammad, W. A., Hajjaji, M. A., Masmali, H. M., Daghreeri, M. J., Alqasmi, A. A., Khormi, A. M., Ghazwani, M. J., Safhi, Y. A. Y. A., & Hakami, A. A. (2026). Harmonization Strategies for Handcrafted Radiomic Features: A Narrative Review of Reproducibility Challenges. Saudi Journal of Health Research & Practice, 2(3 & 4), 26-34. https://doi.org/10.63908/ekk5ja44
Vancouver
Salawi MH, Khawaji KI, Mohammad WA, Hajjaji MA, Masmali HM, Daghreeri MJ, et al. Harmonization Strategies for Handcrafted Radiomic Features: A Narrative Review of Reproducibility Challenges. Saudi Journal of Health Research & Practice. 2026;2(3 & 4):26-34. doi:10.63908/ekk5ja44
BibTeX
@article{Salawi2026,
author = {Salawi, Majed H. and Khawaji, Khalid I. and Mohammad, Waheed A. and Hajjaji, Mohammed A. and Masmali, Hamoud M. and Daghreeri, Mohammed J. and Alqasmi, Ali A. and Khormi, Ahdu M. and Ghazwani, Mohammed J. and Safhi, Yahia A. Yahia A. and Hakami, Amnah A.},
title = {Harmonization Strategies for Handcrafted Radiomic Features: A Narrative Review of Reproducibility Challenges},
journal = {Saudi Journal of Health Research & Practice},
year = {2026},
volume = {2},
number = {3 & 4},
pages = {26--34},
doi = {10.63908/ekk5ja44}
}
EUVIMED