3D CT Texture Analysis In Differential Diagnosis Of Renal Cell Carcinoma

Objective To explore the value of 3D texture analysis technique based on CT images and its related parameters in the differential diagnosis of different pathological types of renal cell carcinoma.Materials and Methods The clinical and imaging data of 73 cases of renal cell carcinoma(54 cases of renal clear cell carcinoma and 19 cases of non-clear cell renal cell carcinoma)without surgery,radiotherapy and chemotherapy and other antitumor treatments were retrospectively analyzed.All patients underwent CT scan and dynamic contrast-enhanced scan.3DQI texture analysis software was used to delineate the boundary of the tumor layer by layer in different scanning periods.The texture features of the entire tumor were collected.We analyzed and compared the difference of texture parameter values(average gray value,standard deviation,skewness,kurtosis,entropy)between different pathological types of renal cell carcinoma,and considered that the difference of texture parameters with P <0.05 was statistically significant.The receiver operating characteristic curve was used to evaluate the diagnostic efficacy of each texture parameter.Results The average gray value,standard deviation and entropy of CT quantitative texture parameters of renal clear cell carcinoma were higher than those of non-clear cell renal carcinoma at different scanning stages,and the kurtosis was lower than that of non-clear cell carcinoma.Mean gray value and standard deviation in the corticomedullary period,parenchymal period,excretory period was significantly different(P <0.05);entropy,kurtosis in unenhanced period,corticomedullary period and parenchymal period difference was statistically significant(P <0.05).There was no significant difference in skewness parameters in different pathological types of renal cell carcinoma at different scanning period(P> 0.05).In the corticomedullary period,the kurt parameters under the area under the ROC curve of 0.87,the sensitivity is 95%,the specificity is 72%,the best threshold is 1.09.In the corticomedullary period,the kurt parameters has high diagnostic value for the identification of renal clear cell carcinoma and renal non-clear cell carcinoma.Conclusion The 3D texture analysis technique based on CT images has certain value in differentiating two different pathological types of renal cell carcinoma and may be a new method to diagnose the pathological type of renal cell carcinoma.