Prediction of lymph node metastasis in patients with papillary thyroid cancer based on radiomics analysis and intraoperative frozen section: a retrospective study

Abstract

Abstract Purpose To evaluate the diagnostic efficiency among clinical model, radiomics model and nomogram that combined radiomics features and frozen section (FS) analysis for the prediction of lymph node (LN) metastasis for the patients with papillary thyroid cancer (PTC). Methods A total of 208 patients with PTC were retrospectively enrolled. The patients were divided into two groups randomly for training groups and validation groups. The Least absolute shrinkage and selection operator (LASSO) regression were used for the selection of radiomics feature extracted from ultrasound (US) images. Univariate and multivariate logistic analysis were used to select predictors including clinical characteristics and FS associated with the status of LN. The clinical model, radiomics model and nomogram were subsequently established. Results Multivariate analysis indicated that age, size group, Adler grade, ACR score and the psammoma body group were independent predictors to predict lymph node metastasis (LNM). The results showed that in the training group, nomogram had better performance than clinical model (P > 0.05) and radiomics model (P < 0.05). In the validation group, the results were similar to the training group, nomogram had a little higher diagnostic efficiency than clinical model (P > 0.05) and radiomic model (P > 0.05). Both in the training and validation group, nomogram had minor non-significant improvements in AUC compared to clinical model and significant improvements compared to radiomic model, however, the sensitivity of nomogram was a little higher. Conclusion We proposed that the nomogram combined the radiomics features and FS had the promise to create a substantial biomarker for predicting LNM of patients with PTC.