Objectives:Due to the different clinical decision making of low grade bladder urotheliral carcinoma and high grade bladder urotheliral carcinoma,We tried to establish a nomogram based on ultrasonic radiomics to predict the pathological grade of bladder tumors,and evaluate the diagnostic efficacy of the nomogram.Methods:From June 2020 to November 2021,we enrolled 191 patients(Median age: 65 years,range: 31-89 years;158 males,33 females)in the Department of Urological Surgical of the First Affiliated Hospital of Nanchang University.According to the proportion of 7: 3,it is randomly divided into training cohorts(n = 133)and validation cohorts(n = 58).After ultrasound image acquisition,image segmentation,preprocessing,849 features were extracted.After dimensionality reduction and feature selection,Rad-Score(Radiomics score)was constructed using the least absolute shrinkage and selection operator(LASSO)algorithm in the training cohorts,whose performance was assessed and then validated in the validation cohorts.Based on the multivariable logistic regression analysis,select significance clinical predictors and then combine the Rad-Score with significance clinical predictors.a radiomics nomogram was developed incorporating the Rad-Score and selected significance clinical predictors.In order to compare,we also created an independent clinical nomogram.Draw receiver operator characteristic curve(ROC)curve,calculate the area under the Curve(AUC),Discrimination and calibration of the nomogram were assessed in the training cohorts and the validation cohorts,respectively.Results:Ten features were selected to calculate Rad-Score.In the training cohorts,high grade bladder urotheliral carcinoma Rad-Score(0.64 [0.47-0.83])was significantly higher than that of the low grade bladder urotheliral carcinoma(0.27 [0.11-0.46])(P<0.001).in the validation cohorts,the high grade bladder urotheliral carcinoma RadScore(0.65 [0.48-0.95])was also significantly higher than that of the low grade bladder urotheliral carcinoma(0.38 [0.17-0.55])(P=0.001).Meanwhile,Rad-Score had excellent diagnostic efficacy,with an AUC of 0.831(95%CI 0.763-0.899)in the training cohorts and 0.822(95%CI 0.717-0.927)in the validation cohorts.The radiomics nomogram developed by combining rad-Score with significance clinical predictors showed good discrimination and calibration.In the training cohorts,radiomics nomogram AUC is 0.901(95% CI 0.850-0.952),which is significantly higher than the clinical nomogram(0.812,95% CI 0.740-0.885)(P = 0.008),The AUC diagnostic capacity of the validation cohorts remained superior(0.864,95% CI 0.770-0.959),which was significantly higher than the clinical nomogram(0.720,95% CI0.585-0.855)(P = 0.012).Conclusions:Ultrasound-based Rad-Score are helpful to identify the pathological grade of bladder tumor before surgery.The performance can be further improved by combining the extracted radiomic features with clinical predictors. |