Use of diffusion-weighted imaging in the noninvasive diagnostic of obstructed biliary ducts.

ABSTRACT

OBJECTIVE: This study sought to evaluate the role of diffusion-weighted imaging (DWI) in differentiation between obstructed and unobstructed bile ducts in patients undergoing magnetic resonance imaging (MRI). METHODS: Eighty-four patients, 40 males and 44 females (mean age 56.4 ± 15.1 years), undergoing MRI with DWI (0-50-500-700) were evaluated and divided into two groups 58 with abnormal laboratory tests (obstructed group) and 26 with normal laboratory values (unobstructed group). Laboratory tests were total bilirubin, alkaline phosphatase, and gamma-glutamyltransferase. Median ADC values were calculated and correlated with laboratory tests and degree of bile-duct dilatation (absent, moderate, or severe). The persistence of signal on DWI (b500 and b700) in the biliary tract was evaluated. Bilirubin values were tested for correlation with bile-duct ADC values and persistence of b700 signal. For statistical analysis, Student t test, chi-square test and Wilcoxon-Mann-Whitney test were used. ADC maps were plotted for three levels of the biliary tree, and a receiver operating characteristic (ROC) curve was calculated. RESULTS: In the obstructed group, 15 patients had severe dilatation, 24 had moderate dilatation, and 19 had no appreciable dilatation; 38 patients had persistent signal on b700 images. In the unobstructed group, 23 patients had no dilatation and 3 had moderate dilatation; 4 patients had persistent signal on b700 images. Correlation was found between degree of bile-duct dilatation, bilirubin levels, persistence of b700 signal, and ADC map values. The calculated ADC map cutoff value (353 10-6 mm2/s) was able to differentiate the obstructed and unobstructed groups with 92.3% sensitivity, 81% specificity, and 91.9% accuracy. CONCLUSIONS: DWI is able to distinguish patients with obstructed versus unobstructed bile ducts, regardless of the degree of dilatation, correlating with clinical and laboratory findings.