Preliminary Study Of Quantitative CT Texture In Differential Diagnosis Of Renal Angiomyolipoma Without Visible Fat And Clear Cell Renal Clear Cell Carcinoma

Objective: To investigate the value of texture features based on enhanced CT examination in cortical and medullary phases in the differential diagnosis of renal angiomyolipoma without visible fat and clear cell renal carcinoma.Methods: 27 cases of renal angiomyolipoma without visible fat and 50 cases of clear cell renal carcinoma in The First Hospital of Lan Zhou University which were confirmed by pathology diagnosis were retrospectively analyzed.The ROI was drawn manually at the images of cortical phase and medullary phase were examined by enhanced CT and 93 CT texture features were extracted by A.K software,Lasso algorithm and multivariate logistic regression were used to preserve the features with P < 0.05,characteristics with the greatest difference were selected in the cortical,medullary and two phases respectively.The area under the receiver operating curve(AUC),sensitivity,specificity,positive predictive value and negative predictive value were used to evaluate the diagnostic efficacy of the model.Finally,Delong test was used to compare the relationship between the AUC of cortical phase,medullary phase and two-phase combined,Chi square test was used to compare the sensitivity and specificity.Results:1)Lasso and multivariate logistic regression were used to reduce the dimension of 93 texture parameters extracted by A.K software twice,two,three and two texture features were reserved in cortical phase,medullary phase and combined phase.2)Logistic regression model was used to establish a texture analysis model to distinguish renal angiomyolipoma and renal clear cell carcinoma,give the result as follows: In cortical phase,AUC of logistic regression model was 0.97(95%CI0.89-0.99),sensitivity 88.9%,specificity 94.1%,positive predictive value 88.9%,negative predictive value 94.1%;The AUC in medullary phase was 0.96(95%CI0.88-0.98),sensitivity 81.5%,specificity 88.2%,positive predictive value 78.6%,negative predictive value 90.0%;The AUC based on the combination of two phases was 0.97(95%CI 0.89-0.99),sensitivity 89.5%,specificity 94.7%,positive predictive value 89.3%,negative predictive value 94.5%.3)There was no significant difference in AUC,sensitivity and specificity among cortical phase,medullary phase and the combination of two phases.Conclusion: The models of CT texture features based on cortical phase,medullary phase and the combination of the two phases have good diagnostic efficacy in the differential diagnosis of renal angiomyolipoma without visible fat and clear cell renal cell carcinoma,this can play a key role in guiding clinicians to choose the treatment plan.