ObjectiveTo evaluate the diagnostic value of texture analysis based on laparoscopic ultrasonography in differentiating renal clear cell carcinoma from renal angiomyolipoma with maximum diameter≤4 cm.MethodsThe data of 124cases of small renal space occupying lesions（maximum diameter≤4 cm

including 66 cases of renal clear cell carcinoma and 58cases of renal angiomyolipoma）confirmed by pathology were retrospectively collected.They were randomly divided into training group（n=87）and verification group（n=37）.Area of interesting was drawn by ITK-SNAP software and the texture features were extracted by python-based Pyradiomics package.Intra-correlation coefficients were used to select the features with excellent stability and repeatability.The minimum-redundancy maximum-relevance（m RMR） and the least absolute shrinkage and selection operator（LASSO）algorithms were used for feature selection and model development for discriminating renal clear cell carcinoma from renal angiomyolipoma.Receiver operating characteristic（ROC） curve analysis was used to evaluate the diagnostic efficacy of the model.ResultsThe best texture feature combination were the glszm

G

ray Level Non Uniformity and the firstorder

9

0 Percentile.The calculation formula for the model was Y=0.185+0.387×glszm

G

ray Level Non Uniformity-0.319×firstorder

9

0 Percentile.The area under ROC curve

sensitivity and specificity were 0.911

0.891

0.854 in the training group and0.900

0.950

0.647 in the verification group

respectively.ConclusionTexture analysis based on laparoscopic ultrasound can accurately identify renal clear cell carcinoma and renal angiomyolipoma with maximum diameter≤4 cm.