Purpose: To investigate the relevance of increasing b-values in evaluating liver fibrosis through the agreement of two diffusion-weighted (DW) sequences.
Materials and methods: A total of 29 cirrhotic patients and 29 healthy volunteers were studied on a 1.5T system. Two single-shot spin-echo echo-planar sequences were acquired using sets of increasing b-values: 0, 150, 250, and 400 seconds/mm(2) (first sequence: DW1a) and 0, 150, 250, 400, 600, and 800 seconds/mm(2) (second sequence: DW2a). Apparent diffusion coefficients (ADCs) of the hepatic parenchyma were calculated on ADC maps. Noise-scaled single-point ADCs were calculated for each sequence from b = 400 seconds/mm(2).
Results: ADCs resulted significantly lower in cirrhotic patients compared to controls using both DW1a (mean 1.14 +/- 0.20 x 10(-3)mm(2)/second vs. 1.54 +/- 0.12 x 10(-3)mm(2)/second; P < 0.0001) and DW2a (mean 0.91 +/- 0.18 x 10(-3)mm(2)/second vs. 1.04 +/- 0.18 x 10(-3)mm(2)/second; P = 0.0089). DW1 and DW2, respectively significantly differed in diagnostic performance at receiver operating characteristic (ROC) curve analysis (P = 0.003), showing AUCs of 0.93 (sensitivity 89.7%, specificity 100%) and 0.73 (sensitivity 62.1%, specificity 79.3%), respectively. Noise-scaled single-point ADCs showed a progressive convergence to similar values in cirrhotic and healthy livers at b = 800 seconds/mm(2) (1.12 +/- 0.27 x 10(-3)mm(2)/second vs. 1.13 +/- 0.17 x 10(-3)mm(2)/second).
Conclusion: A DW sequence is accurate in assessing liver fibrosis using intermediate (400 seconds/mm(2)) rather than high (800 seconds/mm(2)) maximum b-values, but after proper recalculation of ADCs the effects of perfusion rather than diffusion should be considered responsible for the higher accuracy at lower b-values.
(c) 2008 Wiley-Liss, Inc.