Objective To investigate the efficacy and clinical value of using diffusion-weighted imaging(DWI),intra-voxel incoherent motion(I VIM),and diffusion kurtosis imaging(DKI)techniques and their combined parameter models for preoperative prediction of pathological grading in hepatocellular carcinoma(HCC).Methods We prospectively collected imaging data from 70 HCC patients who were confirmed by biopsy or surgical pathology.Preoperative conventional MRI,DWI,IVIM,and DKI scans were performed to obtain various model parameters.Statistical analysis was conducted to com-pare the differences in quantitative parameters among different pathological grades,and correlation analysis,receiver operat-ing characteristic curve analysis,and support vector machine(SVM)prediction model training were performed.Results Significant differences in MK,MD,ADC,and D values were observed between the well-differentiated group and the moder-ately to poorly differentiated group.The parameters of different models showed positive or negative correlations with tissue pathological grading,with MK values negatively correlated and MD,ADC,and D values positively correlated with tumor dif-ferentiation.The diagnostic accuracy,sensitivity,and specificity of the three single models varied among high,moderate,and low-grade HCC,with the DWI+IVIM+DKI combined model demonstrating higher diagnostic efficacy compared to other models,and the IVIM+DKI combined model exhibiting higher predictive accuracy than other models.Conclusion DWI,IVIM,and DKI techniques can be used to predict the pathological grading of HCC.The DWI+IVIM+DKI combined model has higher diagnostic efficacy,while the DKI+IVIM combined model has higher diagnostic accuracy.Multimodal MRI can serve as an effective complement to conventional MRI and assist in the grading diagnosis of hepatocellular carcino-ma.
NETL
NSTL
万方数据
