Reproduce “Radiomics in Medical Imaging: Methods, Applications, and Challenges” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Fnu Neha, Deepak kumar Shukla

This survey provides a comprehensive, methodology-centric analysis of radiomics pipelines, outlining how interdependent design choices from image acquisition to predictive modeling influence feature stability, reproducibility, and clinical translation. It details persistent challenges hindering widespread adoption and proposes future research directions, including hybrid models, multimodal fusion, and standardized benchmarks.

Key insights

  • Minor variations in image acquisition, preprocessing, or segmentation boundaries can substantially alter radiomic feature values, particularly for higher-order texture features, impacting model robustness.
  • Overfitting and validation bias are pervasive in radiomics studies due to high feature dimensionality, extensive model exploration, and insufficient validation rigor (e.g., not performing feature selection within validation loops).
  • A lack of end-to-end standardization in protocols, feature definitions, software, and validation designs across studies severely hinders reproducibility and the accumulation of comparative evidence.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.