Objective To explore the predictive value of spectral CT quantitative parameters combined with radiomics for early liver fibrosis using rat models.Methods Prospective animal experiments were conducted,and the intervention models were constructed for liver fibrosis in rats.A total of 112 spectral CT plain samples were collected from 56 rats,and quantitative parameters(40,60 and 100 keV CT values,slope of the spectral curve)were measured in the spectral CT images to assess the value of each parameter in the staging of liver fibrosis.The regions of interest were sketched out in the 60 keV monoenergetic images using 3D Slicer software,and from which the radiomics features were extracted.The t-test,correlation analysis and the least absolute contraction and selection operator algorithms were used for radiomics feature screening,and the Rad-score was calculated.According to the pathological results,they were classified into non-significant liver fibrosis and significant liver fibrosis,and 3 Logistic regression models(spectral CT model,radiomics model and combined model)were established using the selected spectral CT parameters and radiomics features.The predictive value of these models was evaluated using receiver operating characteristic curves,and calibration curves were plotted to evaluate model fit.Results Except for 100 keV CT value,spectral CT parameters and Rad-score differed statistically between non-significant and significant liver fibroses(P<0.05).The spectral CT model,radiomics model and combined model had AUC of 0.850,0.895,0.939 in the training set,and 0.818,0.803,0.883 in the test set.The calibration curves showed that the 3 models were well fitted,without significant deviation.Conclusion The model constructed with spectral CT and radiomics performes well,worthy of further optimization.

Objective To explore the predictive value of spectral CT quantitative parameters combined with radiomics for early liver fibrosis using rat models.Methods Prospective animal experiments were conducted,and the intervention models were constructed for liver fibrosis in rats.A total of 112 spectral CT plain samples were collected from 56 rats,and quantitative parameters(40,60 and 100 keV CT values,slope of the spectral curve)were measured in the spectral CT images to assess the value of each parameter in the staging of liver fibrosis.The regions of interest were sketched out in the 60 keV monoenergetic images using 3D Slicer software,and from which the radiomics features were extracted.The t-test,correlation analysis and the least absolute contraction and selection operator algorithms were used for radiomics feature screening,and the Rad-score was calculated.According to the pathological results,they were classified into non-significant liver fibrosis and significant liver fibrosis,and 3 Logistic regression models(spectral CT model,radiomics model and combined model)were established using the selected spectral CT parameters and radiomics features.The predictive value of these models was evaluated using receiver operating characteristic curves,and calibration curves were plotted to evaluate model fit.Results Except for 100 keV CT value,spectral CT parameters and Rad-score differed statistically between non-significant and significant liver fibroses(P<0.05).The spectral CT model,radiomics model and combined model had AUC of 0.850,0.895,0.939 in the training set,and 0.818,0.803,0.883 in the test set.The calibration curves showed that the 3 models were well fitted,without significant deviation.Conclusion The model constructed with spectral CT and radiomics performes well,worthy of further optimization.