Effect of radiomics from different virtual monochromatic images in dual-energy spectral CT on the WHO/ISUP classification of clear cell renal cell carcinoma

•WHO/ISUP grading of ccRCC can be distinguished well by monochromatic image radiomics. •Different keV image radiomics has little effect on ccRCC WHO/ISUP grading. •The optimal keV for ccRCC display can be used to sketch the ROI. AIM To investigate the effect of radiomics obtained from different virtual monochromatic images (VMIs) in dual-energy spectral computed tomography (CT) on the World Health Organization/International Association for Urological Pathology (WHO/ISUP) classification of clear cell renal cell carcinoma (ccRCC). MATERIALS AND METHODS A retrospective study of 99 ccRCC patients who underwent contrast-enhanced dual-energy CT was undertaken. ccRCC was confirmed at surgery or biopsy and graded according to the WHO/ISUP pathological grading criteria as low grade (n=68, grade I and II) or high grade (n=31, grade III and IV). Radiomics risk scores (RRSs) for differentiating high and low grades of ccRCC were constructed from 11 sets of VMI in (40–140 keV, 10 keV interval) the cortical phase. Receiver operating characteristic (ROC) curves were drawn and the area under the curves (AUCs) was calculated to evaluate the discriminatory power of RRS for each VMI. The Hosmer–Lemeshow test was used to evaluate the goodness-of-fit of each model and the decision curve was used to analyse its net benefit to patients. RESULTS The AUC values for distinguishing low-from high-grade ccRCC with RRS of 40–140 keV VMIs were all >0.920. The Hosmer–Lemeshow test showed that the p-values of RRS of VMIs were >0.05, suggesting good fits. In the decision curve analysis, RRS from the 40–140 keV VMIs had similar decision curves and provided better net benefits than considering all patients either as high-grade or low-grade. CONCLUSIONS The RRS obtained from multiple VMIs in dual-energy spectral CT have high diagnostic efficiencies for distinguishing between low- and high-grade ccRCC with no significant differences between different VMIs. To investigate the effect of radiomics obtained from different virtual monochromatic images (VMIs) in dual-energy spectral computed tomography (CT) on the World Health Organization/International Association for Urological Pathology (WHO/ISUP) classification of clear cell renal cell carcinoma (ccRCC). A retrospective study of 99 ccRCC patients who underwent contrast-enhanced dual-energy CT was undertaken. ccRCC was confirmed at surgery or biopsy and graded according to the WHO/ISUP pathological grading criteria as low grade (n=68, grade I and II) or high grade (n=31, grade III and IV). Radiomics risk scores (RRSs) for differentiating high and low grades of ccRCC were constructed from 11 sets of VMI in (40–140 keV, 10 keV interval) the cortical phase. Receiver operating characteristic (ROC) curves were drawn and the area under the curves (AUCs) was calculated to evaluate the discriminatory power of RRS for each VMI. The Hosmer–Lemeshow test was used to evaluate the goodness-of-fit of each model and the decision curve was used to analyse its net benefit to patients. The AUC values for distinguishing low-from high-grade ccRCC with RRS of 40–140 keV VMIs were all >0.920. The Hosmer–Lemeshow test showed that the p-values of RRS of VMIs were >0.05, suggesting good fits. In the decision curve analysis, RRS from the 40–140 keV VMIs had similar decision curves and provided better net benefits than considering all patients either as high-grade or low-grade. The RRS obtained from multiple VMIs in dual-energy spectral CT have high diagnostic efficiencies for distinguishing between low- and high-grade ccRCC with no significant differences between different VMIs.