Radiomics Model Based on Contrast-enhanced CT Intratumoral and Peritumoral Features for Predicting Lymphovascular Invasion in Hypopharyngeal Squamous Cell Carcinoma.

RATIONALE AND OBJECTIVES

Patients with Hypopharyngeal Squamous Cell Carcinoma (HSCC) exhibiting lymphovascular invasion (LVI) frequently demonstrate a poor prognosis. We aim to determine whether contrast-enhanced computed tomography (CECT)-derived intratumoral and peritumoral radiomic features could predict the LVI status of HSCC patients.

MATERIALS AND METHODS

166 patients with pathologically confirmed HSCC were included in this study, 47 of whom were LVI positive. Preoperative CECT data were randomly divided into a training dataset and a validation dataset in an 8:2 ratio. A total of 1648 radiomics features were extracted from the total tumor volume (GTV) and the surrounding 1- to 5-mm-wide tumor margins (labeled as Peri1V-5V). A deep learning model based on the GTV was also constructed. Radiomics nomograms were established by integrating deep learning model features and clinical features. Receiver operating characteristic curve (ROC), calibration curve, and decision curve analysis (DCA) were utilized to evaluate and compare the predictive performance of all models.

RESULTS

Peri1V-Radscore showed the best prediction efficiency in the validation dataset among all peritumoral models. Among the clinical variables, the upper tumor boundaries and clinical N stage were independent predictors. Compared with the clinical predictor model, Peri1V-Radscore, deep learn model and Nomogram model can improve prediction efficiency in LVI status. Their respective AUC values were 0.94, 0.84, and 0.96. The results of DCA showed that a good net benefit could be obtained from the Peri1V-Radscore model.

CONCLUSION

Intratumoral combined peritumoral radiomics model based on CECT can superior predict LVI status in HSCC patients and may have significant potential for future applications in clinical practice.