Peritumoral radiomics with clinical nomogram improves differentiation of fat-poor angiomyolipoma from clear cell RCC in challenging cases

Integrating intratumoral and 2mm peritumoral radiomics with clinical data yielded an AUC of 0.829 to differentiate fat-poor angiomyolipoma from clear cell RCC in 194 challenging renal masses, outperforming clinical-only (0.733) and radiomics-only (0.779) models.

• The optimal radiomics model combined intratumoral features with 2mm peritumoral expansion and used XGBoost, achieving AUC 0.779 in the independent test set.
• The nomogram added age and gender, improving AUC to 0.829 in 194 cases with indeterminate imaging or initial imaging-pathology discordance.
• This retrospective study trained on 429 concordant cases, offering a noninvasive tool to potentially reduce unnecessary surgical interventions.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.