To evaluate the diagnostic value of radiomics model based on contrast-enhanced MR Sequences for focal liver lesions
Objective To investigate the value of radiomics model based on contrast-enhanced MR Sequences in the differential diagnosis of focal liver lesions of HCC and FNH.Methods A total of 196 patients with confirmed diagnosis in Zhangzhou Municipal Hospital of Fujian Medical University from 2011 to 2021 were retrospectively analyzed.The radiomics features of contrast-enhanced MR Sequences were extracted,and the variance threshold method,K-best method and LASSO regression operation were used for feature selection and dimensionality reduction.The LR,SVM and KNN classifier models for the differential diagnosis of HCC and FNH were constructed.Results The SVM classifier had the best performance in AUC and sensitivity,the LR classifier had better specificity than other classifiers,and the KNN classifier had the best accuracy.Conclusion The model based on contrast-enhanced MR Imaging sequence shows its advantages in the differential diagnosis of FNH and HCC,but it also has some limitations.The exploration of different classifiers in this study will help to deeply understand the advantages and limitations of each model,so as to improve the accuracy of imaging differential diagnosis of FNH and HCC,and realize individualized precision medicine.
NETL
NSTL
万方数据
