Texture features of MRI in distinguishing hepatic cell carcinoma from focal nodular hy-perplasia
Objective To distinguish hepatic cell carcinoma(HCC)from focal nodular hyperplasia(FNH)based on the tex-ture features of multimodal magnetic resonance imaging(MRI).Methods A retrospective analysis was performed on 38 HCC cases and 14 FNH cases diagnosed pathologically in Hainan Hospital of PLA General Hospital from April 2012 to June 2022.Axial T2 weigh-ted imaging(T2WI)and diffusion weighted imaging(DWI)were used for texture analysis,and the texture features included angular second moment(ASM),contrast,correlation,inverse difference moment(IDM)and entropy.Kolmogorov-Smirnov test and Mann-Whitney text were used to compare the differences of textural feature paraments between the two groups,and multiple regression analy-sis was used to select the variables and their values in the overall model fit,then the regression equation was constructed.Receiver op-erating characteristic curve was used to evaluate the diagnostic efficacy of the regression equation.Results Based on the texture fea-tures of DWI images,it was proved that ASM,correlation and IDM of the FNH group were significantly larger than those in the HCC group,while the contrast and entropy of the FNH group were significantly smaller than those in the HCC group.Based on the texture features of T2 WI images,it was confirmed that ASM and correlation of the FNH group were significantly larger than those in the HCC group,while the contrast,IDM and entropy of the FNH group were significantly smaller than those in the HCC group The selected vari-ables of the DWI group in the overall model fitting were entropy,contrast and constant,and the regression equations Y=-1.621×En-tropy-0.031×Contrast+12.410 was obtained.The selected variables and their values in T2 WI group was the same with the T2 WI com-bined DWI(T2 WI-DWI)group.The selected variables were entropy and constant,and the regression equation Y=-2.595×Entropy+16.419 was obtained.The largest area under the receiver operating characteristic curve was 0.945 for the DWI group.Conclusions Texture features of multimodal MRI can effectively distinguish HCC from FNH.
国家科技期刊平台
NSTL
万方数据
