Correlation Of Histogram Analysis Of Diffusion Kurtosis Magnetic Resonance Imaging With Pathologic Features Of Endometrial Carcinoma:A Comparative Study With Diffusion-weighted Imaging

Part I Quantitative evaluation of diffusion-kurtosis imaging for grading endometrial carcinoma:a comparative study with diffusion-weighted imagingObjective:To evaluate the diagnostic performance of histogram analysis of diffusion kurtosis magnetic resonance imaging?DKI?and standard diffusion weighted imaging?DWI?in discriminating tumor grades of endometrial carcinoma?EC?.Materials and Methods:Seventy-three patients with EC were included in this study.The apparent diffusion coefficient?ADC?value from standard DWI,apparent diffusion for Gaussian distribution(D_app)and apparent kurtosis coefficient(K_app)from DKI were acquired on a 3.0T magnetic resonance imaging?MRI?scanner.The measurement was based on an entire-tumor analysis.Histogram parameters(D_app,K_apppp and ADC)were compared between high-grade?grade 3?and low-grade?grade 1 and 2?tumors.The diagnostic performance of imaging parameters for discriminating high-from low-grade tumors was analyzed by a receiver operating characteristic curve?ROC?.Results:The area under the ROC curve?AUC?of the 10th percentile of D_app,90th percentile of K_app and 10th percentile of ADC were higher than other parameters in distinguishing high-grade tumors from low-grade tumors?AUC=0.821,0.891 and0.801,respectively?.The combination of 10th percentile of D_app and 90th percentile of K_app improved the AUC to 0.901,which was significantly higher than that of the10th percentile of ADC?0.810,p=0.0314?in differentiating high-from low-grade EC.Conclusions:Entire-tumor volume histogram analysis of DKI and standard DWI were feasible for discriminating histological tumor grades of EC.DKI was relatively better than DWI in distinguishing high-grade from low-grade tumor in EC.Part ?Quantitative evaluation of diffusion-kurtosis imaging for depicting myometrial invasion and positive lymphovascular invasion: a comparative study with diffusion-weighted imagingObjective: To investigate the application of histogram analysis of diffusion kurtosis magnetic resonance imaging and diffuson weighted imaging in discriminating deep myometrial invasion from superficial myometiral invasion and positive lymphovascular invasion?LVSI+?from negative lymphovascular invasion?LVSI-?of endometrial carcinoma?EC?.Materials and Methods: A total of 118 patients pathologically confirmed to have EC underwent preoperative diffusion-weighted magnetic resonance imaging?b value of 0,700,1400,2000 s/mm2?3.0T were included in this study from January 2016 to December 2017 in our department.The data was postprocessed with a bio-exponential diffusion kurtosis imaging?DKI?model for quantitation of apparent diffusion values?Dapp?and apparent kurtosis coefficient values?Kapp?for non-Gaussian distribution.The data was postprocessed with monoexponential diffusion weighted imaging?DWI?model for quantiation of apparent diffusion coefficience?ADC?.The measurements were based on an entire-tumor analysis approach.Histogram parameters of Dapp,Kapp and ADC were compared between deep myometrial invasion and superficial myometrial invasion,positive lymphovacular invasion and negative lymphovascular invasion subgroups using independent t test or Mann-Whitney U test.The diagnostic performance of the imaging parameters were assessed for discrimination of the presense of deep myometrial invasion and positive lymphovascular invasion using receiver operating characteristic regression?ROC?.The presence of deep myometial invasion and lymphatic metastasis between positive and negative lymphovascular subgroups were compared using the Chi-square test or Fisher exact test.The tumor volume between two subgroups was compared using the independent t test.Results: Patients with deep myometrial invasion had significantly lower Dapp values in mean,10 th percentile,90 th percentile,higher Kapp values in mean and 90 th percentile,as well as lower ADC values in mean,10 th percentile,90 th percentile than those of superficial myometrial invasion?p<0.05 for all?.The skewness and kurtosis of Dapp and ADC were higher than those of superficial myometrial invasion subgroups?p<0.05 for all?.ROC analysis revealed that 10 th percentile of Dapp,mean of Kapp and the 10 th percentile of ADC had better diagnostic performance in distinguishing deep myometrial invasion with relatively higher AUC?0.724,0.658,and 0.710,respectively?when compared with other parameters.The diagnostic performance of 10 th percentile of Dapp seems to be higher than that of 10 th percentile of ADC?0.724 vs 0.710?,the difference did not reach the statistical significance?p=0.351?.Patients with positive lymphovascular invasion had significantly lower Dapp values in mean,10 th percentile,90 th percentile,higher Kapp values in 90 th percentile,as well as lower ADC values in mean,10 th percentile,90 th percentile than those of negative lymphovascular invasion?p<0.05 for all?.The skewness of Dapp and ADC were higher than those of negative lymphovascular invasion?p<0.05 for all?.ROC analysis revealed that mean of Dapp,90 th percentile of Kapp and the mean of ADC had better diagnostic performance in distinguishing positive lymphovascular invasion with relatively higher AUC?0.778,0.653,and 0.743,respectively?when compared with other parameters.The diagnostic performance of the mean of Dapp seems to be higher than that of mean of ADC?0.778 vs 0.743?,the difference did not reach the statistical significance?p=0.224?.The mean tumor volume of deep myometrial invasion and positive lymphovascular invasion subgroups were significantly larger than those of superficial myometrial invasion and negative lymphovascular subgroups?p=0.009,0.039?.The presence of deep myometrial invasion and lymphatic metastasis in positive lymphovascular subgroup were significantly higher than those of negative lymphovascular invasion subgroup?p<0.001 and p=0.004?.Conclusions: Distribution of DKI and conventional DWI parameters characterized by histogram analysis may be useful in distinguishing deep myometrial invasion and positive lymphovascular invasion of endometrial carcinoma.The convential monoexponential DWI modes yielded as least the equivalent diagnostic performances in discriminating deep myometrial invasion and positive lymphovascular invasion to the bioexponential DKI model.The positive lymphovascular invasion have more prevalence in deep myometrial invasion and lymphatic metastasis.