CT Radiomics to Differentiate Between Wilms Tumor and Clear Cell Sarcoma of the Kidney in Children

Abstract

Abstract Objective: To find out the role of contrast-enhanced CT radiomics in distinguishing pediatric Wilms' tumor (WT) from clear cell sarcoma of the kidney (CCSK). Materials and Procudres: We retrospectively gathered the clinical information and preoperative CT images of 83 children with WT and 33 children with CCSK. In the renal parenchymal phase of contrast-enhanced CT, the maximum tumor diameter, the ratio of the maximum CT value of the solid portion of the tumor to the mean CT value of the contralateral renal vein (CTmax/CT renal vein), and the dilated peritumoral cysts were analyzed. To extract radiomics features from arterial phase CT images, all patients were randomly divided into a training set (n=81) and a test set (n=35) in the ratio of 7:3.Sampling Technique (SMOTE) was used to handle imbalanced datasets, these radiomics features were then filtered using Pearson correlation coefficient and LASSO algorithm, and the filtered features were built into a machine learning based classifier model to calculate receiver operating characteristic curve (ROC), area under the ROC curve (AUC), 95% confidence interval (CI), accuracy, sensitivity and specificity. Results: While there was no statistically significant difference between WT and CCSK in the test set (P>0.05), there was a statistical difference between the maximum tumor diameter (p=0.021) and dilated peritumoral cyst (p=0.005) in the training set. Nine radiomics features were used to develop the radiomics model, and machine learning based on logistic regression was chosen to build it.The cross-validation AUC was 0.888 (95% CI: 0.804-0.972), accuracy was 0.864, sensitivity was 0.826, and specificity was 0.879. The AUC for the test set was 0.784 (95% CI: 0.604-0.964), accuracy was 0.829, sensitivity was 0.600, and specificity was 0.920.The AUC for the training set was 0.901 (95% CI: 0.826-0.976), accuracy 0.889, sensitivity 0.826. Conclusion: Radiomics of contrast-enhanced CT images is of diagnostic value in analyzing and differentiating WT and CCSK in children.