Diagnostic performance of strain ratio combined with conventional ultrasound for cervical lymph node metastasis in head and neck neoplasms

Cervical lymph node metastasis is a key prognostic factor in head and neck neoplasms, yet conventional ultrasound alone is limited by morphological overlap between metastatic and benign nodes. This study aimed to evaluate the diagnostic performance of a combined model integrating the ultrasound strain elastography (USE)-derived strain ratio (SR) with conventional ultrasound parameters and to assess its superiority over individual parameters alone. This prospective diagnostic-accuracy study with blinded index-test consecutively enrolled patients diagnosed with head and neck neoplasms who had cervical lymph node biopsies at the Affiliated Hospital of Xuzhou Medical University between January and December 2025. They were categorized into a metastatic group (N₁) and a non-metastatic group (N₀) based on pathological results. The study compared conventional ultrasound parameters, long-to-short axis ratio (L/S), lymph node hilar architecture, and color Doppler flow imaging (CDFI) classification and SR between the two groups. A binary logistic regression model was developed to incorporate all four variables, and a nomogram was constructed for individualized prediction. Restricted cubic spline (RCS) was used to assess the correlation of SR and cervical lymph node metastasis. Diagnostic performance was assessed and compared using receiver operating characteristic (ROC) curve analysis and DeLong’s test for area under the curve (AUC) comparisons. Model calibration and clinical utility were further assessed using a calibration curve and decision curve analysis (DCA). Subgroup analyses stratified by sex, age, and education level were also performed. 90 patients with head and neck neoplasms were analyzed. Four parameters (age, sex, marital status, and smoking) differed significantly between the N1 and N0 groups (all P < 0.05). RCS analysis result revealed that SR had a significant association with cervical lymph node metastasis ( P -overall = 0.005). Among individual parameters, CDFI classification achieved the highest AUC of 0.809. The combined model demonstrated superior diagnostic performance, with an AUC of 0.910 (95% CI: 0.831–0.960), sensitivity of 89.5%, specificity of 78.8%, and accuracy of 83.3%, significantly outperforming each individual parameter by DeLong’s test (all P < 0.05). The calibration curve indicated acceptable model fit (Hosmer-Lemeshow P = 0.549), and decision curve analysis confirmed positive net benefit across a wide range of threshold probabilities (0-0.90). Subgroup analyses revealed that combined model had favorable clinical utility and wide generalizability (all AUC > 0.870). The combined model integrating SR with conventional ultrasound parameters achieved an AUC of 0.910, significantly outperforming each individual parameter, and demonstrated favorable calibration and clinical utility. These findings support its potential as a practical and radiation-free tool for preoperative cervical lymph node evaluation in patients with head and neck neoplasms, warranting further validation in multicenter studies.