| Purpose: To investigate the value of clinical-radiomics model to predict the postoperative stone-free rate(SFR)of flexible ureteroscopy lithotripsy(FURL)for kidney stones patients.Patients and methods: This study retrospectively analyzed 2129 kidney stones patients treated with FURL.All subjects underwent non-contrast enhanced computed tomography(NCCT)scans before surgery.At the same time,relevant clinical information of each patient was collected.The patients were divided into successful group and failure group according to SFR results three months after operation.142 patients with lower calyx calculi were included in further radiomics analysis.The procedure can be divided into five parts: 1)Data preprocessing,3D Slicer,an image segmentation software,was applied to delineate region of interest(ROI)layer by layer and fuse them into volume of interest(VOI);2)Features Extraction,604 texture features are automatically extracted by the MATLAB;3)99 patients were included in the training set according to the ratio of 7:3 between the training set and verification set.Data dimension reduction and feature selection were carried out by the least absolute shrinkage and selection operator(LASSO)regression analysis using R software.The linear equation of the selected features was adopted to calculate the radiomics score for each patient;4)Multivariate logistic regression analysis was employed to build a model incorporating radiomics and potential clinical factors,and the nomogram was drawn;5)The remaining 43 patients were used for model verification.Regarding the performance evaluation of the model,the receiver operating characteristic(ROC)curve was used to measure the model discrimination,the calibration curve was applied to evaluate the calibration,and the decision curve was employed to quantify clinical application value.Results: 264 patients with single kidney stones were finally recruited in the study. The overall SFR of FURL was about 72%,whereas the SFR for lower calyx calculi and non-lower calyx calculi was 56.3% and 90.16%,respectively.The independent predictors of SFR were radiomics score,stone volume,operator’s experience,and hydronephrosis level,all of which were included in the model.The area under the curve(AUC)of this model was 0.949 and 0.947 in primary and validation cohort,respectively.Moreover,the results of calibration curve showed good agreement between prediction by nomogram and actual observation.and decision curve shows that the model has high clinical application value.Conclusion: Radiomics based on kidney NCCT images can better predict SFR of kidney stone after FURL.The clinical-radiomics model obtained by combining relevant clinical indicators can quantitatively evaluate the success rate of surgery.Purpose: To explore the value of radiomics in assessing the risk factors of sepsis after flexible ureteroscopy lithotripsy(FURL)or percutaneous nephrolithotomy(PCNL)in patients with ureteral calculi.Patients and methods: Patients who underwent FURL or PCNL treatment from March 2013 to December 2018 were retrospectively studied.Relevant clinical information was collected and all participants preoperatively underwent non-contrast enhanced computed tomography scans.After propensity score matching(PSM),the radiomics model construction and evaluation process is carried out according to the following four steps:1)Image segmentation,using 3D Slicer software to delineate the region of interest(ROI)layer by layer,and merge it into a volume of interest(VOI);2)Feature extraction,texture features are automatically extracted by the 3D Slicer internal texture extraction plug-in;3)Feature selection and model construction,using R language statistical analysis software.The Least Absolute Shrinkage and Selection Operator(LASSO)regression analysis method was carried out for image data dimensionality reduction and feature selection.The constructed model is used to calculate the radiomics score of each patient;4)Model performance evaluation,ROC(Receiver operating characteristic)curve was used for the discrimination estimation,calibration curve for the calibration evaluation,and decision curve for the evaluation of clinical application value.Results: Totally,847 patients with single ureteral calculi were involved in the study.The overall incidence of sepsis after FURL or PCNL was 5.9%.The sepsis group and the non-sepsis group were statistically different in the following indicators(all p values <0.001): gender,preoperative fever,urine culture,albumin,globulin,albumin and globulin ratio,urine white blood cells(qualitative),urine white blood cells(quantitative),urine nitrite.After PSM,132 patients were included for radiomics model construction.884 radiomics features were extracted by 3D Slicer software.After LASSO analysis,the λ value that provides the best data fit for the model is0.0202,and the final model incorporated 26 key parameters.The AUC of this model is 0.881(95% CI,0.813-0.931),the sensitivity is 79.55%,and the specificity is96.59%.Meanwhile,the decision curve analysis shows that the model has satisfactory clinical application value.Conclusion: Radiomics model can effectively predict the risk factors of sepsis in patients with single ureteral calculi after undergoing FURL or PCNL surgery,and help early identify high-risk patients with sepsis. |
