Prediction of preoperative the Stone-Free rate in percutaneous nephrolithotomy based on CT clinical-radiomics nomogram: a comparative study with Guy’s stone score and S.T.O.N.E score.

Abstract

Abstract Purpose: This study aimed to develop a clinical-radiomics nomogram by combining clinical factors and radiomics features.The objective of the nomogram was to predict the stone-free rate(SFR) in percutaneous nephrolithotomy (PCNL). Additionally, the predictive performance of the nomogram was compared with Guy's stone score (GSS) and S.T.O.N.E score. Patients and methods: A retrospective analysis was conducted on 109 suitable patients with solitary kidney stones who underwent PCNL at the Second Affiliated Hospital of Nanchang University from April 2021 to October 2022.The preoperative clinical data and non-contrast-enhanced CT images of all patients were collected.Radiomic features were extracted from the CT images after preprocessing steps such as wavelet transformation, logization, and resampling. The least absolute shrinkage and selection operator (LASSO) method was used to screen the radiomics features and calculate the radiomics score for each patient using lasso regression coefficient.Multivariate logistic regression analysis was performed to determine the clinical independent predictive factors. These factors were combined with radiomics to construct a clinical-radiomics model, which was visualized using a nomogram.The GSS and S.T.O.N.E score of each patient were calculated and analyzed. The predictive performance of the clinical-radiomics nomogram,Guy’s stone score (GSS),and S.T.O.N.E score was compared and analyzed through identification,calibration,and clinical benefit. Results: The postoperative statistics revealed a stone-free rate of 65.1%. The results from multivariate logistic regression analysis indicated that the number and diameter of stones were independent risk factors for residual stones after percutaneous nephrolithotomy (PCNL). In the training set, the clinical-radiomics nomogram, GSS and S.T.O.N.E score showed area under the receiver operating curve (ROC) values of 0.925, 0.772 and 0.712, respectively. In the validation set, the AUCs for the clinical-radiomics nomogram, GSS and S.T.O.N.E score were 0.944, 0.786 and 0.714, respectively.The Delong test demonstrated that the clinical-radiomics nomogram exhibited better discrimination ability than GSS and S.T.O.N.E score (p<0.05).The calibration curve and Hosmer-Lemeshow (HL) test confirmed the good calibration ability and fit of the clinical-radiomics nomogram.Furthermore, the decision curve analysis (DCA) revealed that the clinical-radiomics nomogram provided a better net benefit compared to GSS and S.T.O.N.E score. Conclusion:The clinical-radiomics nomogram constructed based on clinical characteristics and radiomics features can well predict the stone-free rate after PCNL, and its predictive performance is better than the GSS and S.T.O.N.E score.