Objective:To assess the predictive value of intravoxel incoherent motion(IVIM)and diffusion kurtosis imaging(DKI)in determining molecular subtypes expression in breast cancer.Methods:A total of 244 patients of breast cancer confirmed by pathology were collected,and they were divided into 4 molecular subtypes,Luminal A type,Luminal B type,HER-2-positive type and triple-negative type.IVIM and DKI scanning were performed in all patients.The values of true diffusion coefficient(D),perfusion-related diffusion coefficient(D*),perfusion fraction(f),mean kurtosis(MK)and mean diffusivity(MD)were analyzed.The ROC curve was drawn to compare the diagnostic efficiency.Results:Among the 244 cases,32 cases were Luminal A type,131 cases were Luminal B type,36 cases were HER-2-positive type and 45 cases were triple-negative type.The D*value of Luminal A type was lower than that of the other three types(P<0.01).The f value of triple-negative type was higher than that of the other three types(P<0.01).Luminal B type had a lower MD value and a higher MK value than the HER-2-positive type and triple-negative type(all P<0.05).The D,D*,f,MD and MK values were more valuable for distinguishing between Luminal A/B type and non-Luminal type,and the D* and MD values had the larger AUCs.Combining the IVIM and DKI models,the AUC was 0.752,the diagnostic efficacy was higher than that of a single parameter(Z=1.845～4.144,P=0.001～0.045).The D,D*,MD and MK values were valuable for identifying HER-2-positive type and other molecular subtypes,and the D and MD values had the large AUCs.The D*,f and MD values were valuable for identifying triple-negative type and non-triple-negative type,and the sensitivity and accuracy of the combined three parameters were higher than those of a single parameter.Conclusion:IVIM combined with DKI can be used to predict the molecular subtype in breast cancer,which is beneficial for the individualized treatment.
Breast neoplasmsMagnetic resonance imagingIntravoxel incoherent motionDiffusion kurtosis imaging
国家科技期刊平台
NSTL
维普
万方数据
