CT-Based Radiomics in the Characterization of Solid Renal Tumors: A Systematic Review.
Authors
Affiliations (3)
Affiliations (3)
- Department of Clinical Radiology, University Hospital of Ioannina, University Campus, 45110 Ioannina, Greece.
- Department of Neurosurgery, Faculty of Medicine, School of Health Sciences, University of Ioannina, University Campus, 45110 Ioannina, Greece.
- Department of Clinical Radiology, Faculty of Medicine, School of Health Sciences, University of Ioannina, University Campus, 45110 Ioannina, Greece.
Abstract
<b>Background:</b> Renal cell carcinoma (RCC) is a global health challenge characterized by significant histological heterogeneity. Conventional contrast-enhanced CT often struggles to differentiate RCC from solid benign renal tumors like fat-poor angiomyolipoma (fpAML) and renal oncocytoma (RO), leading to potential surgical overtreatment. CT-based radiomics has emerged as a promising non-invasive approach that extracts high-dimensional quantitative imaging features to support lesion characterization and may contribute toward more comprehensive, biopsy-adjacent decision support, although it does not yet replace histopathological assessment. <b>Methods:</b> This review systematically evaluates the predictive performance of CT-based radiomics in characterizing solid renal tumors. A literature search was conducted in PubMed/MEDLINE, Cochrane, and Scopus databases for original research published between 2012 and 2025. The review focuses on four key areas: differentiating benign renal tumors from RCC, clear cell (ccRCC) from non-ccRCC, fpAML from RCC, and RO from RCC. <b>Results:</b> In total, 47 studies were assessed, including 11,999 patients. CT-based radiomics demonstrates high diagnostic performance across all categories. Median Area Under the Curve values were 0.830 (0.747-0.900) for benign vs. malignant differentiation, 0.900 (0.861-0.910) for ccRCC vs. non-ccRCC, 0.912 (0.879-0.933) for fpAML vs. RCC, and 0.885 (0.841-0.947) for RO vs. RCC. The integration of radiomic features with clinical parameters into combined nomograms consistently yielded the highest predictive accuracy. <b>Conclusions:</b> Radiomics provides a non-invasive, objective method to characterize renal tumors, potentially reducing unnecessary surgeries and enabling personalized treatment. However, widespread clinical adoption remains limited by a lack of protocol standardization, the need for automated segmentation, and the requirement for prospective, multicenter validation.