Feasibility study of radiomics combined with CT and clinical features for early diagnosis of PTC-CLNM

Abstract

Abstract Background: Papillary thyroid carcinoma (PTC) is an indolent tumor, but central lymph node metastasis (CLNM) occurs at an early stage. Early diagnosis of papillary thyroid carcinoma central lymph node metastases (PTC-CLNM) is very important for patient prognosis. So, the purpose of this study is to construct a multi-dimensional early diagnosis model by combining traditional computed tomography (CT) imaging features, clinical features and radiomics features, so as to improve the early diagnosis ability of PTC-CLNM and improve the treatment effect of PTC. Methods: A total of 226 patients with PTC who underwent head and neck or thyroid enhanced CT examinations in Zhejiang Provincial People's Hospital from January 2021 to February 2022 were included in this study. The patients were randomly divided into training set (n=180) and validation set (n=46). Imaging histologic features of individual patient were derived from pre-operative plain scan, enhancement scan arterial phase and intravenous phase images. Radiomics and multi-dimensional models were constructed using support vector machine. The properties of the multi-dimensional model were evaluated using receiver operating characteristics (ROC) on the training and test sets, and its utility for clinical purposes was assessed by Decision Curve Analysis (DCA). Results: A total of 930 radiomics features were extracted from the three-phase CT images of each patient, from which 8 features related to CLNM were filtered. Four clinical factors (sex, age, and long and short diameters of tumors) were significantly associated with CLNM. The areas under the ROC curves for the training and validation sets in the multi-dimensional model were 0.870 (95% confidence interval [CI] = 0.818-0.921) and 0.819 (95% CI = 0.681-0.956), respectively. Decision curve analysis showed that the multidimensional model had better clinical utility than the other models. Conclusion: The multi-dimensional radiomics model combined with traditional CT imaging features, clinical risk features and radiomics features is meaningful toward early diagnosis of Central neck node Metastasis in patients with PTC.