Multiphase dynamic contrast-enhanced magnetic resonance imaging radiomics nomogram for preoperative differentiation of sinonasal inverted papilloma-associated squamous cell carcinoma from benign inverted papilloma
Background Reliable preoperative distinction between sinonasal inverted papilloma–associated squamous cell carcinoma (SNIP-SCC) and benign sinonasal inverted papilloma (SNIP) is important for the planning of treatment. Methods This retrospective study analyzed 304 patients who were pathologically diagnosed with SNIP or SNIP-SCC from two centers. Among them, 214 patients from Center A formed the development cohort, in which model performance was internally evaluated using 20 repetitions of stratified nested five-fold cross-validation; 90 patients from Center B served as the external validation cohort. Radiomics feature descriptors were derived from conventional magnetic resonance imaging (MRI) and dynamic contrast-enhanced MRI in the early, peak/plateau, and delayed phases. Results Increasing age, convoluted cerebriform pattern mutation, and a washout-type time–signal intensity curve were independently associated with SNIP-SCC. The multiphase contrast-enhanced radiomics model achieved an internally validated area under the receiver operating characteristic curve (AUC) of 0.870 [95% confidence interval (CI), 0.813–0.921] in the development cohort and an AUC of 0.843 (95% CI, 0.735–0.932) in the external validation cohort. A nomogram integrating the radiomics score with the selected clinical and imaging predictors yielded AUCs of 0.899 (95% CI, 0.853–0.941) and 0.883 (95% CI, 0.755–0.979), with corresponding accuracies of 84.1% and 85.6%, respectively. Conclusion Multiphase dynamic contrast-enhanced MRI radiomics provides complementary information to conventional clinical and imaging assessment and may facilitate preoperative differentiation of SNIP-SCC from benign SNIP.
Authors
- XU Jin-e
- Jin Zhou
- Lin-Sha Yang
- Ying Guo
- Duo Zhang
- Yan Wu
- Rui-Chao Liu
- Chong Liu
Institutions
- First Hospital of Qinhuangdao (CN)
- Baoding People's Hospital (CN)
Publication Details
- Journal
- Frontiers in Medicine
- Published
- 2026-11-03
- DOI
- https://doi.org/10.3389/fmed.2026.1980539
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00