Is Water Diffusion Isotropic in the Cirrhotic Liver? A Study with Diffusion-weighted Imaging at 3.0 Tesla Rossano Girometti, MD, Gennaro Esposito, PhD, Daniele Bagatto, MD, Claudio Avellini, MD, Massimo Bazzocchi, MD, Chiara Zuiani, MD Rationale and Objectives: Water molecule motion has been shown to be isotropic in healthy liver as assessed by diffusion-weighted imaging (DWI). The purpose of this study was to investigate whether this assumption is still valid in the cirrhotic liver. Materials and Methods: Twelve cirrhotic patients and 12 controls underwent DWI on a 3.0T-system. We used an echo-planar sequence independently applying unidirectional motion-probing gradients along read-(x), phase-(y), and slice-(z) directions, respectively. Liver apparent diffusion coefficient (ADC) and perfusion fraction (f) were calculated along each gradient direction, based on two b-values sets (0–400 and 0–800 seconds/mm 2 ). Measurement was repeated at two different axial slices of the liver. Results: No significant difference among x, y, and z- ADC (and f) of the liver was found within controls and cirrhotic patients (P > .01), regardless of the set of b-values or the slice of measurement. ADC was lower in cirrhotic patients than in controls (difference of 0.24–0.39  10 3 mm 2 /seconds at b = 800 seconds/mm 2 ; P = .000–.0139). Perfusion fraction f was lower in cirrhotics than in controls, irrespective of the b-values set or the slice of measurement (difference of 0.05–0.15; P < .0001–.0885). Conclusion: The liver shows isotropic water diffusion in cirrhotics, despite fibrotic distortion leading to decreased ADC as compared to controls. Our results emphasize that the correlation between parenchymal changes in liver fibrosis and the ADC estimate remains an elusive goal based on the state-of-the-art DWI technique. Key Words: Liver cirrhosis; liver fibrosis; effusion-weighted imaging; liver isotropy. ªAUR, 2012 O ver the last years, diffusion-weighted Imaging (DWI) has been investigated as a tool to provide noninvasive detection (1) and quantification of liver fibrosis (2,3). Fibrosis results from the accumulation of extracellular matrix components, which causes distortion of the parenchymal architecture (4), and leads to the restriction of water diffusion (5). As expected, the hepatic apparent diffusion coefficient (ADC), measured by means of DWI, has been shown to decrease proportionally to the degree of fibrosis and/or inflammation (6–9). Nonetheless, the mechanism underlying the decrease of the ADC is not completely understood, and probably depends more on alterations in liver perfusion rather than in actual diffusion (5,10,11). Thus, the radiopathologic correlation of what DWI measures in liver fibrosis must be further elucidated. In addition, several concerns still limit DWI clinical applica- tions for liver fibrosis. Among them, technical issues like the choice of b-values, sequence design and ADC assessment opti- mization are those mainly affecting the reproducibility of results (4). In this scenario, the most appropriate direction of DWI motion-probing gradients is still undetermined. In previous studies, gradients have been applied multi- (1,7,9,12) or mono-directionally, based on the assumption that the liver shows isotropic water diffusion (5,6,13). To our knowledge, this assumption has been investigated on humans only by Taouli et al on a 1.5-T system (14). No previous studies were aimed to assess liver isotropy at higher magnetic field strength, and whether isotropy is lost (or not) in liver fibrosis. On this basis, the purpose of our study was to investigate the pattern of water diffusion (ie, isotropic versus nonisotropic) in fibrotic livers as compared to controls, by using a DWI sequence at 3.0T. Acad Radiol 2012; 19:55–61 From the Institute of Diagnostic Radiology (R.G., D.B., M.B., C.Z.), Institute of Pathology (C.A.), Az. Ospedaliero-Universitaria S.Maria della Misercordia, University of Udine, via Colugna n. 50, 33100 Udine, Italy and Department of Biomedical Sciences and Technologies and MATI Centre of Excellence (G.E.), University of Udine - P.le Kolbe n. 4, 33100 Udine, Italy. Received August 3, 2011; accepted September 7, 2011. Address correspondence to: R.G. e-mail: rgirometti@sirm.org ªAUR, 2012 doi:10.1016/j.acra.2011.09.009 55