Radiomics analysis of CT imaging improves preoperative prediction of cervical lymph node metastasis in laryngeal squamous cell carcinoma.

Abstract

To investigate the role of CT radiomics for preoperative prediction of lymph node metastasis (LNM) in laryngeal squamous cell carcinoma (LSCC). LSCC patients who received open surgery and lymphadenectomy were enrolled and randomized into primary and validation cohorts at a ratio of 7:3 (325 vs. 139). In the primary cohort, we extracted radiomics features from whole intratumoral regions on venous-phase CT images and constructed a radiomics signature by least absolute shrinkage and selection operator (LASSO) regression. A radiomics model incorporating the radiomic signature and independent clinical factors was established via multivariable logistic regression and presented as a nomogram. Nomogram performance was compared with a clinical model and traditional CT report with respect to its discrimination and clinical usefulness. The radiomics nomogram was internally tested in an independent validation cohort. The radiomics signature, composed of 9 stable features, was associated with LNM in both the primary and validation cohorts (both p < .001). A radiomics model incorporating independent predictors of LNM (the radiomics signature, tumor subsite, and CT report) showed significantly better discrimination of nodal status than either the clinical model or the CT report in the primary cohort (AUC 0.91 vs. 0.84 vs. 0.68) and validation cohort (AUC 0.89 vs. 0.83 vs. 0.70). Decision curve analysis confirmed that the radiomics nomogram was superior to the clinical model and traditional CT report. The CT-based radiomics nomogram may improve preoperative identification of nodal status and help in clinical decision-making in LSCC. • The radiomics model showed favorable performance for predicting LN metastasis in LSCC patients. • The radiomics model may help in clinical decision-making and define patient subsets benefiting most from neck treatment.