CT-Based Radiomics in Renal Tumors: Current Evidence, Methodological Challenges, and Future Perspectives for Precision Oncology

Background: CT-based radiomics may provide non-invasive imaging biomarkers for renal-tumor characterization and risk stratification. This systematic review evaluates the clinical evidence and translational readiness of radiomics in renal oncology and reports an exploratory methodological appraisal of a clearly delimited subgroup. Methods: PubMed/MEDLINE and Embase were searched from database inception through 15 August 2026 using controlled vocabulary and free-text terms for renal tumors, computed tomography, and radiomics or quantitative image analysis. Eligibility was restricted to English-language original studies published from 1 January 2020 to 15 August 2026 that were available in full text. Engineered-feature radiomics constituted the primary evidence base; end-to-end deep-learning studies were considered separately. Owing to clinical and methodological heterogeneity, findings were synthesized narratively. Results: The searches retrieved 1521 records (PubMed/MEDLINE, n = 357; Embase, n = 1164). The PubMed/MEDLINE stream yielded 242 unique records after removal of 115 duplicates; 72 full-text reports were assessed and included. Cross-deduplication and screening of the Embase records identified no additional eligible study. Contemporary evidence supports potential applications in benign–malignant differentiation, histological subtype classification, WHO/ISUP grade and pathological-stage prediction, and postoperative outcome assessment. Reported discrimination was frequently high, but performance estimates were not directly comparable and often declined in independent testing. Within the illustrative, non-representative 28-report detailed appraisal subset, 25 engineered-radiomics studies had an available numerical RQS (median, 16; interquartile range, 15–20; range, 12–24; 44.4% of the maximum). Conclusions: CT radiomics is technically promising but not yet ready for routine clinical use. Quality-related findings apply only to the assessed subgroup and cannot characterize the entire evidence base. Standardized acquisition and feature definitions, transparent analysis, clinically relevant comparators, prospective multicenter validation, formal risk-of-bias assessment in future reviews, and impact studies are needed.