A Scoring System Based on Diffusion Tensor Imaging and Blood Biochemistry Tests for Diagnosing Biliary Atresia in Infants

TL;DR

This study developed a noninvasive scoring system using MRI and blood tests to help diagnose biliary atresia in infants with high accuracy.

Contribution

A novel scoring system combining diffusion tensor imaging and blood biochemistry for diagnosing biliary atresia in infants.

Findings

The scoring system achieved 95.65% accuracy in the validation cohort for predicting biliary atresia.

The system showed 100% sensitivity and 88.89% specificity in the validation cohort.

Apparent diffusion coefficient values and gamma glutamyl transpeptidase levels were key factors in the scoring system.

Abstract

Objectives: The aim of this study was to investigate the diagnostic value of a scoring system based on diffusion tensor imaging (DTI) and blood biochemistry tests for biliary atresia (BA) in infants. Methods: Seventy-four patients who had undergone DTI and blood biochemistry tests were included in this study. Among them, 51 (36 BA patients and 15 non-BA patients) were assigned to the training cohort, and 23 (14 BA patients and 9 non-BA patients) were assigned to the validation cohort. The characteristics that significantly differed between the groups in the training cohort were used to develop a scoring system for predicting the presence or absence of BA through binary logistic regression analysis. The scoring system was subsequently validated in the validation cohort, and its diagnostic performance was assessed with receiver operating characteristic curve analysis. Results: The mean apparent diffusion coefficient values of the hepatic right and caudate lobes and the serum levels of gamma glutamyl transpeptidase were selected for constructing the scoring system. The accuracy, sensitivity, and specificity of the system in predicting BA were 82.35%, 91.67% and 60%, respectively, in the training cohort and 95.65%, 100% and 88.89%, respectively, in the validation cohort. The areas under the receiver operating characteristic curve in the training cohort and validation cohort for predicting BA were 0.87 and 0.94 (p ≤ 0.001 each), respectively. Conclusions: We developed a relatively noninvasive scoring system for diagnosing BA according to the results of DTI and blood biochemistry tests, which demonstrated good performance and may be a potential method for differentiating BA in infants.

Figures