Abstract:
Objective: To construct a scoring system based on multiparametric magnetic resonance imaging (MRI) and blood biochemical parameters, and to evaluate its value in the diagnosis of biliary atresia (BA).
Methods: This study included 79 neonates and infants confirmed with BA or non-BA, all of whom underwent MRI and blood biochemical tests. The MRI examinations included three-dimensional magnetic resonance cholangiopancreatography (3D-MRCP) and diffusion tensor imaging (DTI). 3D-MRCP assessed the development of the extrahepatic bile ducts and gallbladder, while DTI measured the apparent diffusion coefficient (ADC) and fractional anisotropy values of the right, caudate, and left hepatic lobes. The blood biochemical tests measured levels of total bilirubin, direct bilirubin, indirect bilirubin, aspartate aminotransferase, alanine aminotransferase, gamma-glutamyl transpeptidase (GGT), and platelet count. Abnormal development of the extrahepatic bile ducts and gallbladder displayed by 3D-MRCP, along with the mean ADC values of the right and caudate lobes measured by DTI and the average serum GGT level, were used as feature parameters to construct the diagnostic model for BA using a binary logistic regression classifier. The model parameters were weighted and fused, and the predicted values were subjected to logit transformation to generate a single diagnostic index (ranging from 0 to 1) for predicting whether an individual patient has BA. When this index was greater than 0.5, it indicated the possibility of BA. Intraoperative cholangiography, liver biopsy, and clinical treatment outcomes were used as the gold standards to assess the diagnostic performance of the fused diagnostic index, including accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and the area under the receiver operating characteristic curve (AUC).
Results: The performance of the scoring system in predicting BA was as follows: accuracy was 86.08%, sensitivity was 90.57%, specificity was 76.92%, positive predictive value was 88.89%, negative predictive value was 80%, and AUC was 0.9 (P<0.001).
Conclusion: Based on MRI data and blood biochemical parameters, a relatively non-invasive scoring system for BA was developed. This scoring system demonstrated good diagnostic performance and has the potential to serve as an effective means to distinguish whether neonates and infants have BA.
