The Value Of Clinical Factors And CT Radiomics In Predicting Cervical Lymph Node Metastasis Of Papillary Thyroid Carcinoma

Objective:To investigate the value of clinical factors and CT Radiomics in predicting cervical lymph node metastasis(CLNM)of papillary thyroid carcinoma(PTC).Methods:1.A total of 262 patients with pathologically confirmed PTC from May 2015 to October 2022 in the First Affiliated Hospital of Jinan University were enrolled,and they were divided into the training group(183 cases)and the validation group(79 cases)according to the ratio of 7:3.All patients underwent standardized enhanced CT scan before surgery,and the CT images and clinical data of the patients were collected.Based on CT arterial phase images,3D-Slicer(a texture analysis software)was used to delineate contour of PTC layer by layer,and Radiomics features were extracted.Lasso regression was used to screen CLNM-related features.Mann-whitney U test or Chi-square test(Χ~2test)were used to identify clinical variables associated with CLNM.2.The statistically significant Radiomics features and clinical variables were included in the binary Logistic regression analysis,then the Clinical model,CT radiomics model and Clinical-CT Radiomics model were constructed,the receiver operating Characteristic curve(ROC)analysis of the Logistic regression model was constructed to evaluate its diagnostic efficacy,and verified in the validation group.Results:1.The Clinical-CT Radiomics model(AUC=0.804,sensitivity=68.7%,specificity=82.4%)showed the highest efficacy of predicting lymph node metastasis than the CT Radiomics model(AUC=0.683,sensitivity=65.2%,specificity=60.3%)and the Clinical model(AUC=0.782,Sensitivity=69.6%,specificity=79.4%),whose diagnostic differential efficacy was validated in the validation group.2.In the validation group,the Clinical-CT Radiomics model(AUC=0.782,sensitivity=84.4%,specificity=61.8%)was more effective in predicting lymph node metastasis than the CT Radiomics model(AUC=0.628,sensitivity=75.6%,Specificity=58.8%)and Clinical model(AUC=0.768,sensitivity=84.4%,specificity=64.7%).Conclusions:Clinical-CT Radiomics model combining Radiomics features achieved the good results in predicting CLN metastasis of papillary thyroid carcinoma.