The application of nomogram model integrating clinical factors and multi-modal MRI radiomics features for predicting cervical lymph nodes metastasis for patients with oral tongue squamous cell carcinoma: a multicenter study

Abstract

Background The aim of this study was to build and test the nomogram model integrating clinical factors and multi-modal MRI radiomics features to predict the risk of cervical lymph nodes metastasis (CLNM) in oral tongue squamous cell carcinoma (OTSCC) patients, and further validate the clinical utility of the model in cT1-T2 subgroup and cN0 subgroup. Materials and Methods A total of 326 patients enrolled from 2 centers made up group Ⅰ (including T1-Weighted Images (T1WI) and Fat-suppressed T2-Weighted Images (FS-T2WI)) and group Ⅱ (including T1WI, FS-T2WI and T1-Weighted Contrast Enhanced MRI (CE-MRI)). Patients were divided into training set, internal test set and external test set. Radiomics features from the three MRI sequences and clinical factors were extracted and selected. Radiomics models, clinical models and the combined models (nomogram models) integrating clinical factors and multi-modal MRI radiomics features were constructed respectively in two groups. cN0 subgroup and cT1-T2 subgroup analyses were performed. The performance of the three models was compared. Results The combined model integrating clinical factors and multi-modal MRI radiomics features in group Ⅱ had the best performance, with the AUCs of 0.983, 0.944 and 0.770 in the training set, internal test set and external test set. cN0 subgroup and cT1-T2 subgroup analyses confirmed excellent performance of the combined model in group Ⅱ, with the AUCs of 0.975 (0.944–0.992), 0.944 (0.720-1.000) and 0.728 (0.553–0.843) for cN0 subgroup, and the AUCs of 0.959 (0.860-1.000), 0.960 (0.871-1.000) and 0.804 (0.506-1.000) for cT1-T2 subgroup. Conclusions The combined model based on clinical factors and multi-modal MRI radiomics features including T1WI, FS-T2WI and CE-MRI can well predict the risk of preoperative CLNM for OTSCC patients.