AUTHOR=Yu Pengyi , Wu Xinxin , Li Jingjing , Mao Ning , Zhang Haicheng , Zheng Guibin , Han Xiao , Dong Luchao , Che Kaili , Wang Qinglin , Li Guan , Mou Yakui , Song Xicheng 

TITLE=Extrathyroidal Extension Prediction of Papillary Thyroid Cancer With Computed Tomography Based Radiomics Nomogram: A Multicenter Study

JOURNAL=Frontiers in Endocrinology

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2022.874396

DOI=10.3389/fendo.2022.874396

ISSN=1664-2392

ABSTRACT=Objectives: To develop and validate a CT based radiomics nomogram for preoperative predicting of ETE in PTC patients
Methods: A total of 153 patients were randomly assigned to training and internal test sets (7:3). 46 patients were recruited to serve as an external test set. A radiologist with 8 years of experience segmented the images. Radiomics features were extracted from each image and Delta-radiomics features were calculated. Features were selected by using a one-way analysis of variance and the least absolute shrinkage and selection operator in the training set. K-nearest neighbor, logistic regression, decision tree, linear-support vector machine (linear -SVM), gaussian-SVM, and polynomial-SVM were used to build 6 radiomics models. Next, a radiomics signature score (Rad-score) was constructed by using the linear combination of selected features weighted by their corresponding coefficients. Finally, a nomogram was constructed combining the clinical risk factors with Rad-scores. Receiver operating characteristic (ROC) curve, decision curve analysis (DCA), and calibration curve were performed on the three sets to evaluate the nomogram’s performance.
Results: 4 radiomics features were selected. The six models showed a certain value of radiomics, with AUCs from 0.642 to 0.701. The nomogram combining the Rad-score and clinical risk factors (radiologists’ interpretation) showed good performance (internal test set: AUC 0.750; external test set: AUC 0.797). The calibration curve and DCA demonstrated good performance of the nomogram. 
Conclusion: Our radiomics nomogram incorporating the radiomics and radiologists’ interpretation has utility in the identification of ETE in PTC patients.