Classification Analysis Of Patients With Covert Hepatic Encephalopathy Based On Precunnes Radiomic Features

Objective: The significant abnormalities of precuneus(PC),which is associated with brain dysfunction,have been identified in cirrhotic patients with covert hepatic encephalopathy(CHE).The present study aimed to apply radiomics analysis to identify the significant features in PC and their subregions of HBV-related cirrhotic patients,then to build and evaluate classification models for CHE.Methods: From February 2018 to June 2019,106 HBV-related cirrhotic patients(54 CHE patients and 52 patients without hepatic encephalopathy)were included from the second affiliated hospital of Chongqing medical university.All the patients underwent the Psychometric Hepatic Encephalopathy Score(PHES)examination,and a3.0T magnetic resonance high-resolution 3D T1 WI scan of the skull was used to obtain three-dimensional T1-weighted images(3D-T1WI).For each participant,PC and their subregions were segmented and extracted a large number of radiomic features,then identified the features withsignificant discriminative power.Multivariable logistic regression models were constructed and evaluated on the basis of identified features and clinical risk factors.Furthermore,a radiomics nomogram was developed based on the model with the best classification performance for individual prediction of CHE.Results: A large number of radiomic features were extracted from the PC and their subregions,and then 4 best radiomic features was selected as follows: right Medial area 7(PEp)_variance HLH,right Area31(Lc1)_median HLL,right Lc1_GrayLevelNonuniformity(GLN)LLL and right Lc1_Informational Measure of Correlation 1(IMC1)LLL.The combined model based on radiomics signature and clinical risk factors achieved best classification performance.Testing set independent tested that the receiver operating curve(ROC)area under the curve(AUC),sensitivity and specificity is 0.926,100.0% and 76.5%,respectively.And the radiomics nomogram based on the combined model has a good calibration,could be used for individualized prediction of CHE.Conclusion: This study presented that the radiomic features of PC and its subregions,especially the right PC subregion,may be a potential imaging-marker of CHE.The radiomics nomogram that incorporates the radiomics signature and clinical risk factors may facilitate the individualized prediction of CHE.