The Value Of Magnetic Resonance Image Texture Analysis For The Subtype Of Renal Cell Carcinoma And Pathological Grade Of Clear Cell Renal Cell Carcinoma

Part one The Diagnostic Value of Magnetic Resonance Image Texture Analysis for the Subtype of Renal Cell CarcinomaObjective To investigate the value of MR image(FS-T2WI)texture analysis in differentiating different subtypes of renal cell carcinoma.Method Retrospective analysis of 31 patients with ccRCC(15 low stage[stages ?and ?]and 16 high stage[stages ? and ?])and 13 patients with chRCC and 11 patients with pRCC,all RCC cases having been confirmed by surgery and pathological examination,and all patients underwent routine MRI examination.Using full-volume method to outline the region of interest(ROI)on FS-T2WI,and collecting the texture analysis quantitative parameters:Mean,Entropy,Standard Deviation(StDev),Inhomogeneity,Skewness and Kurtosis.The independent sample t-test was used to compare the differences in texture analysis parameters between ccRCC and non-ccRCC,low-stage ccRCC and chRCC,low-stage ccRCC and pRCC,high-stage ccRCC and chRCC,high-stage ccRCC and pRCC,chRCC and pRCC.Logistic regression analysis was used to perform multi-parameter joint analysis,and the diagnostic performance of each parameter was calculated using the receiver operating characteristic curve(ROC).Result The Entropy,StDev and Inhomogenity between ccRCC and non-ccRCC were statistically significant(Entropy:t=3.753,P?0.000;StDev:t=3.489,P?0.001;Inhomogenity:t=-2.722,P?0.009).The Entropy,Mean and StDev between low-stage ccRCC and chRCC were statistically significant(Entropy:t=2.387,P?0.008;Mean:t=2.932,P?0.007;StDev:t=3.836,P?0.001).The Entropy,Mean,StDev,Kurtosis and Inhomogenity between low-stage ccRCC and pRCC were statistically significant(Entropy:t=2.809,P?0.01;Mean:t=4.008,P?0.001;StDev:t=4.193,P?0.000;Kurtosis:t=-2.345,P?0.028;Inhomogenity:t=2.991,P?0.006).The Entropy and Mean between high-stage ccRCC and chRCC were statistically significant(Entropy:t=2.435,P?0.022;Mean:t=-2.056,P?0.05).The Entropy between high-stage ccRCC and pRCC were statistically significant(Entropy:t=2.467,P?0.021).There are no statistically significant between chRCC and pRCC.Entropy showed statistical significance in the identification of renal cell carcinoma of each group.Combined with multi-parameter analysis and receiver operating characteristic curve(ROC),the efficacy of identifying subtypes of renal cell carcinoma in each group was improved.Conclusion MR image(FS-T2WI)texture analysis can help to identify different subtypes of renal cell carcinoma,providing a reference for the preliminary diagnosis and preoperative prediction of renal cell carcinoma.Part two The Value of Magnetic Resonance Image Texture Analysis in Differentiating Low-and High-Stage Clear Cell Renal Cell CarcinomaObjective To evaluate the value of MR image(FS-T2WI)texture analysis in pathological grading of clear cell renal cell carcinoma(ccRCC).Method 31 cases with clear cell renal cell carcinoma(15 low stage[stages ? and ?]and 16 high stage[stages ? and ?])confirmed by surgery and pathology were retrospectively analyzed,all patients underwent MRI examination.Using full-volume method to outline the region of interest(ROI)on FS-T2WI,and collecting the texture analysis quantitative parameters:Mean,Entropy,Standard deviation(StDev),Inhomogeneity,Skewness and kurtosis.The independent sample t-test was used to compare the differences in texture analysis parameters between the low-and high-stage ccRCC.Logistic regression analysis was used to perform multi-parameter joint analysis,and the diagnostic performance of each parameter was calculated using the receiver operating characteristic curve(ROC).Result Mean,StDev and Kurtosis had statistical significance between low-and high-stage ccRCC:Mean(t=4.833,P?0.05);StDev(t=3.816,P?0.05);Kurtosis(t=-2.843,P?0.05).When using Mean to distinguish low-and high-stage ccRCC,its AUC,associated criterion,sensitivity,specificity respectively is 0.925,298.075,93.75%,93.33%;when using StDev to distinguish low-and high-stage ccRCC,its AUC,associated criterion,sensitivity,specificity respectively is 0.846,91.55,100%,60%;when using Kurtosis to distinguish low-and high-stage ccRCC,its AUC,associated criterion,sensitivity,specificity respectively is 0.742,0.05,93.75%,53.33%.The multi-parameter joint analysis was used to identify the efficiency of low-and high-stage ccRCC.When using Mean and StDev to distinguish low-and high-stage ccRCC,its AUC,sensitivity,specificity respectively is 0.921,100%,86.67%;when using Mean and Kurtosis to distinguish low-and high-stage ccRCC,its AUC,sensitivity,specificity respectively is 0.917,93.75%,93.33%;when using StDev and Kurtosis to distinguish low-and high-stage ccRCC,its AUC,sensitivity,specificity respectively is 0.879,93.75%,73.33%;when using Mean,StDev and Kurtosis to distinguish low-and high-stage ccRCC,its AUC,sensitivity,specificity respectively is 0.929,87.50%,93.33%.Conclusion Part of quantitative parameters(Mean,StDev,Kurtosis)of texture analysis of MR image(FS-T2WI)can help to identify low-and high-stage ccRCC,which provides reference for clinicians to make treatment plan and prognosis evaluation.