Objective:To investigate the value of enhanced MRI radiomics to preoperatively predict Ki-67 expression in hepatocellular carcinoma(HCC).Methods:The clinical and enhanced MRI data of 101 HCC patients were retrospectively analyzed.A threshold of 10%was used to classify the Ki-67 high expression(62 cases)and low expression(39 cases).Radiomics features were extracted from the tumor regions of enhanced MRI at the arterial phase(AP),portal venous phase(PVP),and delayed phase(DP).Single-phase and three-phase combined radiomics logistic regression models were constructed using five-fold cross-validation in the training cohort(80 cases).Combined with the clinical data and image features of the patients,a clinical-imaging-radiomics combined model was established.The diagnostic efficacy of each model was compared by ROC curve.Finally,the validation cohort(21 cases)was verified,and the best diagnostic model was selected.Results:The maximum diameter and the edge of the tumor were independent risk factors for the high expression of Ki-67.The clinical-imaging regression model showed that the AUCs of the diagnostic efficacy of the training cohort and the validation cohort were 0.764 and 0.759,respectively.The AUCs of the AP,PVP,DP and three-phase combined model of enhanced MRI in the training cohort were 0.819,0.804,0.865,0.915,and those in the validation cohort were 0.730,0.720,0.807,0.843,respectively.The AUCs of the clinical-imaging-radiomics combined model for the training cohort and the validation cohort were 0.924,0.880,respectively.Conclusion:The clinical-imaging-radiomics combined model based on enhanced MRI can provide a great value in preoperative prediction for Ki-67 expression in HCC.
Magnetic resonance imagingRadiomicsKi-67Carcinoma,hepatocellular
国家科技期刊平台
NSTL
万方数据
