Reproducibility in medical image radiomic studies: contribution of dynamic histogram binning

TL;DR

This paper investigates how dynamic histogram binning affects the reproducibility of radiomic features in medical imaging, highlighting its sensitivity to annotation fluctuations and proposing static binning as a potential remedy.

Contribution

It identifies the impact of dynamic histogram binning on radiomic feature stability and advocates for static binning to improve reproducibility in radiomic studies.

Findings

Dynamic binning increases sensitivity to annotation fluctuations.

Static binning can mitigate reproducibility issues.

Highlights need for standardized binning methods in radiomics.

Abstract

The de facto standard of dynamic histogram binning for radiomic feature extraction leads to an elevated sensitivity to fluctuations in annotated regions. This may impact the majority of radiomic studies published recently and contribute to issues regarding poor reproducibility of radiomic-based machine learning that has led to significant efforts for data harmonization; however, we believe the issues highlighted here are comparatively neglected, but often remedied by choosing static binning. The field of radiomics has improved through the development of community standards and open-source libraries such as PyRadiomics. But differences in image acquisition, systematic differences between observers' annotations, and preprocessing steps still pose challenges. These can change the distribution of voxels altering extracted features and can be exacerbated with dynamic binning.