Multimodal CT radiomics and inflammatory marker fusion model improves preoperative parotid mass risk stratification

Internally validated multimodal CT radiomics and inflammatory marker model achieved AUC 0.931 (sensitivity 0.957, specificity 0.782) for preoperative parotid mass risk stratification in 490 patients. Exploratory: AUC 0.952 vs MSRSGC 0.712.

• Multimodal model AUC 0.931 (95% CI 0.888–0.973), sensitivity 0.957, specificity 0.782 in internal validation.
• In exploratory FNAB subset, model outperformed MSRSGC (AUC 0.952 vs 0.712; p<0.001; NRI 0.399, IDI 0.386).
• Preliminary findings require prospective multicenter validation.

Automated summary

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