| Abstract:Objective：Differential diagnosis of retroperitoneal dedifferentiated and well differentiated liposarcoma using DWIADC image texture analysis. Methods：The patients with retroperitoneal liposarcoma confirmed by operation and pathology in Zhongshan Hospital, Fudan University from November 2009 to December 2018 were retrospectively analyzed, among which 36 cases were dedifferentiated liposarcoma, and 81 cases were highly differentiated liposarcoma, all patients were examined by abdominal or pelvic MR plain scan, DWI and multiphase enhanced scan. Texture analysis was carried out by using Siemens multiparametric software of Germany. Gray histogram parameters such as Skewness, E. Kurtosis, DIffEntropy, Contrst Entropy, ADC mean, ADC median, 5% ADC value, 95% ADC value, and ADC trivalue (low, medium, and high) were obtained. ShapiroWilk test was used to test the normal distribution of the data, which accorded with the normal distribution. Independent sample ttest was used to compare the data of the nonnormal distribution. MannWhitney U test was used to test the data of the nonnormal distribution. The ROC curve of the variables that were different between the two groups was used to evaluate the differential diagnostic efficacy. Results：There were 117 patients. There was significant difference in the Volume, Mean, Median, P95, Skewness, and E.Kurtosis between the two groups (P<0.05). The area under ROC curve of E.Kurtosis, Skewness, Mean, Median, P95, and Volume was about 0.665, 0.738, 0.635, 0.633, 0.652, and 0.819, respectively. The sensitivity was about 63.89%, 61.11%, 83.33%, 82.86%, 62.27%, and 97.22%, and the specificity was about 77.78%, 86.42%, 53.75%, 51.22%, 66.71%, and 59.26%. E.Kurosis, Skewness, Mean, Median, P95, and Volume can be used to distinguish the two groups of cases. Conclusions：DWIADC image texture analysis is helpful in the differential diagnosis of retroperitoneal dedifferentiation and welldifferentiated liposarcoma, among which Skewness, Kurtosis, and Volume are more effective. |