Whole-Liver Apparent Diffusion Coefficient Histogram Analysis For Staging Of Liver Fibrosis

Objective:To discuss the research value of whole‐liver ADC histogram parameters in the pathologic grading of liver fibrosis.Methods:From January 2017 to January 2019,totally twenty individuals with no liver disease and 86 patients with liver fibrosis,including 30 with chronic viral hepatitis,29 with autoimmune hepatitis,and 27 with unexplained liver fibrosis patients who underwent diffusion-weighted MR imaging（b values of 0 and 800s/mm²）were included in our hospital.A region of interest（ROI）was drawn in each slice of the axial diffusion‐weighted images with dedicated software.The 3D whole‐liver ROI ADC value histogram and histogram parameters were obtained with dedicated software by accumulating all ROIs in each slice.Mann–Whitney U‐test was firstly applied in four groups for comparisons of all the ADC value histogram parameters between each stage,and then a ROC curve was constructed to evaluate the effectiveness of ADC value histogram parameters with significant differences in predicting stage 1 or greater（≥F1）,stage 2 or greater（≥F2）,stage 3 or greater（≥F3）liver fibrosis.The diagnostic performance（specificity,sensitivity,negative likelihood ratio,positive likelihood ratio,negative predictive value,and positive predictive value）of the best ADC value histogram parameters（with the highest area under ROC curve of each group）for identifying different fibrosis stages of different groups and the corresponding optimal cutoff values were calculated.Results:1.In the viral hepatitis group,the liver fibrosis stage was F1 in six patients,F2 in six patients,F3 in eight patients,and F4 in ten patients.In the autoimmune hepatitis group,the liver fibrosis stage was F1 in six patients,F2 in seven patients,F3in eleven patients,and F4 in five patients.In the unexplained liver fibrosis group,the liver fibrosis stage was F1 in five patients,F2 in nine patients,F3 in six patients,and F4 in seven patients.2.Kurtosis,entropy,skewness,mode,and 90^thand 75^thpercentiles exhibited significant differences among the pathological fibrosis stages（P?<?0.05）.In the combined liver fibrosis group,kurtosis achieved the highest AUC（0.801;95%confidence interval[CI]:0.702–0.900;sensitivity:0.750;specificity:0.850;positive likelihood ratio:4.953;negative likelihood ratio:0.302;positive predictive value:0.946;negative predictive value:0.486）,with a cutoff value of 1.817,in differentiating fibrosis stage≥F1.3.In autoimmune hepatitis group,the best diagnostic indicator in predicting fibrosis stage≥F3 was entropy（AUC=0.764;95%CI:0.545,0.983）.And beyond that,kurtosis was found to be the most meaningful parameter in differentiating all fibrosis stages of the viral hepatitis,autoimmune hepatitis,unexplained liver fibrosis group and combined liver fibrosis group.In viral hepatitis group,the highest AUC were0.752（95%CI:0.603,0.902）,0.793（95%CI:0.656,0.931）and 0.768（95%CI:0.628,0.909）respectively in predicting fibrosis stage≥F1,≥F2 and≥F3.In autoimmune hepatitis group,the highest AUC were 0.771（95%CI:0.623,0.919）and0.739（95%CI:0.584,0.894）respectively in predicting fibrosis stage≥F1 and≥F2.In unexplained liver fibrosis group,the highest AUC were 0.798（95%CI:0.662,0.934）and 0.722（95%CI:0.566,0.878）respectively in predicting fibrosis stage≥F1 and≥F2.In combined liver fibrosis group,the highest AUC were 0.801（95%CI:0.702,0.900）and 0.703（95%CI:0.586,0.819）respectively in predicting fibrosis stage≥F1and≥F2.There was no significant indicator in discriminating fibrosis stage≥F3 in the unexplained liver fibrosis and combined liver fibrosis group.Conclusion:1.The significant ADC histogram parameters in identifying different fibrosis stages of different groups are not all the same.2.Kurtosis,entropy,skewness,mode,and 90^thand 75^thpercentiles may contribute to the staging of liver fibrosis,especially kurtosis.