Histopathological correlations of CT-based radiomics imaging biomarkers in native kidney biopsy

Choi et al. evaluated the correlation between CT-based radiomic features and chronic histological changes in native kidney biopsies. They analyzed data from adults who had undergone non-contrast CT scans and kidney biopsies within a week. Using Aview Research software, the left kidney was segmented, and various radiomic features, including shape, first-order, and high-order texture features, were extracted. Their machine learning model, based on texture features from the whole kidney parenchyma, predicted moderate-to-severe fibrosis with an AUC of 0.89. The findings suggest that radiomic texture features could reflect chronic histological changes, potentially offering a non-invasive alternative to kidney biopsy. This approach could be especially useful in patients with contraindications for biopsy. However, the study’s retrospective, single-center design limits broader applicability. Future studies should focus on multi-center validation and consider integrating other clinical data for enhanced prediction models, aiming to support non-invasive diagnostics for kidney fibrosis.