Magnetic Resonance Imaging Radiomics Study Of Hepatic Alveolar Echinococcosis With Multiple Organ Metastasis

Objective: To develop and validate the MRI radiomics model for predicting extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis,and to explore its predictive value for extrahepatic multiple organ metastasis in patients with hepatic alveolar echinococcosis.Methods: 117 patients with hepatic alveolar echinococcosis(69 cases with extrahepatic multiple organ metastasis and 48 cases withoutextrahepatic multiple organ metastasis)confirmed by postoperative pathology or imaging in our hospital from January 2015 to October 2019 were analyzed retrospectively.They were randomly divided into training cohort(n = 83)and test cohort(n = 34)in the ratio of7:3.The clinical baseline indexes and MRI features of hepatic alveolar echinococcosis were collected,and the clinical model was established by univariate and multivariate analysis.Draw the region of interest(ROI)from T2WI-SPAIR and contrast-enhanced T1WI(T1-C)delayed phase images respectively.Extract radiomics features in ROI using Py Radiomics software.Through univariate rank sum test,correlation analysis and multivariate stepwise regression,the radiomics features closely related to the extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis were selected.According to the above features,the radiomics models of T2WI-SPAIR and T1-C were constructed respectively,and the optimal MRI model for predicting the multi organ metastasis of hepatic alveolar echinococcosis was selected.The optimal MRI radiomics model and clinical model were fused to establish a clinical radiomics combined model.The nomogram presents the clinical radiomics combined model in a visual form.The receiver operating(ROC)curve,the area under the curve(AUC),sensitivity,accuracy,and specificity were drawn in the training and the test cohorts respectively to evaluate the predictive performance of the constructed model for extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis.The accuracy recall(PR)curve and Delong test were used to compare the AUC values between different prediction models.The clinical application value of multiple prediction models was compared using a decision analysis curve(DCA).Calibration curves are drawn to evaluate the accuracy of different prediction models.Results: 1.1037 radiomics features were extracted from T2WI-SPAIR and T1-C(delayed phase)images respectively,of which 6 radiomics features were closely related to extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis.2.Compared with T2WI-SPAIR model,T1-C radiomics model has higher prediction performance,the clinical radiomics combined model is the most effective in predicting extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis.The AUC value,sensitivity,accuracy and specificity of the training cohort of clinical radiomics combined model were 0.861,0.898,0.819 and 0.706 respectively,and the AUC value,sensitivity,accuracy and specificity of the test cohort were 0.868,0.950,0.765 and 0.500 respectively.3.In training cohort and test cohort,there was no significant difference in AUC between T1-C model and the clinical radiomics combined model(P>0.05).4.The results of calibration curve and decision curve analysis in training and test cohorts show that the clinical radiomics combined model has the supreme dependability.5.CA125 level and hepatic vein involvement were independent predictors of extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis.The AUC of the clinical model established according to the above factors in the training cohort and test cohort were 0.739 and 0.777 respectively.Conclusions: The radiomics model based on T1-C sequence has good accuracy in predicting extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis.The nomogram combined with the features of radiomics and clinical risk factors can further improve the prediction efficiency of extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis.To some extent,MRI radiomics can provide clinical support for the diagnosis of extrahepatic multiple organ metastasis of hepatic alveolar echinococcosis,help to formulate effective personalized treatment strategies for patients with hepatic alveolar echinococcosis,and improve the survival and treatment of patients.