In a preliminary study, we showed that the 8 CT parameter variation pairwise group comparisons had statistically significant differences on average in 78/86 radiomics features. On the other hand, 84% of the univariate radiomics feature tests had a successful and statistically significant differentiation of the 4 classes of liver tissue. We concluded that the differences in radiomics feature values obtained from different types of liver tissue are generally greater than the intraclass differences resulting from CT parameter variations.

The phantom was printed using the Phantom X solution detailed in the study, "Radiopaque Three-dimensional Printing: A Method to Create Realistic CT Phantoms". It relies on inkjet cartridges filled with potassium iodide solutions (600 mg/mL) with prints realized on plain paper (80 g/m2). Stacked paper sheets result in three-dimensional phantoms.  Because the printing process is very specific to the Phantom X company and no other printer can be used, we did not make the printer creation file publicly available.

This dataset can serve as a reference to assess both the discriminative power and stability of radiomics features with CT variations. In addition, it can be used to develop efficient data harmonization techniques to improve robustness of (deep) radiomics features and models.  Additional information and source code related to the project can be found at https://qa4iqi.github.io/.