DWI、IVIM和DKI定量参数联合模型在HCC病理分级中的应用评估

Application and Evaluation of Combined DWI,IVIM,and DKI Quantitative Parameter Model in Pathological Grading of HCC

目的探讨肝细胞癌(HCC)术前利用扩散加权成像(DWI)、体素内不相干运动(IVIM)和扩散峰度成像(DKI)技术及其相关参数联合模型对病理分级的预测效果和应用价值。方法前瞻性地搜集了70例HCC患者穿刺活检或手术病理确诊的影像资料,术前行常规MRI、DWI、IVIM、DKI扫描,获得各模型参数。通过统计分析比较不同病理分级间各模型定量参数的差异,并进行相关性分析、受试者工作曲线分析、支持向量机(SVM)预测模型训练。结果高分化组与中低分化组之间的MK、MD、ADC和D值之间的差异显著。不同模型中各参数与组织病理学分级呈正负相关,其中MK值与肿瘤分化度呈负相关,MD、ADC、D值与肿瘤分化程度呈正相关。三种单一模型各参数在高、中、低分化HCC的诊断准确性、敏感性和特异性方面均有差异,但DWI+IVIM+DKI三联模型诊断效能高于其他模型、IVIM+DKI二联模型预测准确性高于其他模型。结论DWI、IVIM和DKI技术可用于HCC病理分化程度的预测诊断,其中DWI+IVIM+DKI联合模型具有更高的诊断效能,DKI+IVIM联合模型具备更高的诊断准确率。多模态MRI可作为常规MRI的有效补充,辅助HCC分级诊断。

Objective To investigate the efficacy and clinical value of using diffusion⁃weighted imaging(DWI),intra⁃voxel incoherent motion(IVIM),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 compare the differences in quantitative parameters among different pathological grades,and correlation analysis,receiver operating 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 moderately 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 differentiation.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 carcinoma.