The quantitative evaluation of hepatic function is important for monitoring patients and preoperative assessment of the hepatic reserve1. In clinical practice, tests that assess global liver function may fail to detect regional liver dysfunction. Thus, liver function may be estimated more accurately by using imaging-based tests, which can detect both regional and global liver function. Intravoxel incoherent motion (IVIM) diffusion-weighted imaging (DWI) is a promising method for the assessment of diffuse liver disease, given its potential for providing multi-parametric information and combinations of diffusion and perfusion effects. Chronic liver diseases are associated with extracellular matrix accumulation, which may affect both true diffusion and microcirculation. Therefor this work aims to assess the sensitivity of IVIM in evaluating liver function in patients with chronic liver disease.Eighty patients (57 men, 23 women; mean age, 54 ± 12 years) who had chronic liver disease and had undergone IVIM DWI using 10 different b values (0, 10, 30, 60, 100, 150, 200, 400, 600 and 1000 s/mm2) at 3.0 T MR scanner (Achieva, Philips Healthcare, The Netherlands) were included. IVIM DWI was performed by using a respiratory triggered single-shot echo planar imaging (EPI factor, 53) sequence with a parallel imaging SENSE(SENSE factor, 2))( TR/TE, 1642/62 ms; matrix, 256; FOV, 375 × 302 × 176 mm; slice thickness, 5 mm; slice gap, 0.5 mm; slices, 32; NSA, 2). Patients were subdivided into the following three groups 2: patients with MELD scores ≤ 10 (n = 32), 11–18 (n = 23) and > 18 (n = 25). All curve-fitting algorithms were created using a home-developed program based on MATLAB (MathWorks, Natick, MA) (Figure 1, 2). Pure molecular diffusion (D), pseudo-diffusion (D*), perfusion fraction (f) and apparent diffusion coefficient (ADC) values were calculated, and used to evaluate liver function, as indicated by the MELD score. Intraclass correlation coefficients (ICCs) were calculated to determine the consistency between the data gathered by the two radiologists. One-way analysis of variance was used to evaluate IVIM parameters in patients with different grades of liver function. Receiver operating characteristic (ROC) curve analysis was used to determine diagnostic accuracyThe ICC values of D, D*, f and ADC between the two radiologists were 0.997, 0.986, 0.985 and 0.995, respectively. D*, f and ADC values significantly decreased with increasing MELD scores (p < 0.001)(Table1, 2, Figure 3). MELD scores were inversely correlated with D*, f and ADC values (r = -0.672, r = -0.605, r = -0.538, respectively). The areas under the curve (AUCs) of D* and f for evaluating liver function were 0.73–093 and 0.82–0.95, respectively (Table 3).Our results showed generally good agreement of each parameter between the two radiologists, even for the D* and f parameters, suggesting that the use of whole-liver data extraction may increase the stability of the data. In our study, true molecular diffusion coefficients (D values) were weakly correlated with MELD scores. This suggests that the association of reduced ADC values with advanced liver disease merely reflects decreased perfusion in the microvessels rather than restricted molecular diffusion in the tissue. In chronic liver disease, liver fibrosis and elevated intrahepatic resistance lead to a reduction in portal blood flow, the increase in hepatic arterial blood flow is insufficient to fully offset the decrease in portal blood flow, which leads to a reduction in D* values. Moreover, hepatic steatosis is often observed in the impaired liver parenchyma, and has been associated with decreased hepatic parenchymal perfusion. Both D* and f values are perfusion-related parameters, however, they pertain to different characteristics of perfusion. D* values reflect endovascular blood flow velocity, while f values reflect vascular volume3. Our results demonstrated that f values significantly differed between different stages of liver function, and the correlation of f values with MELD scores was stronger than that of D* and ADC values with MELD scores. Thus, f values may be a sensitive indicator for liver function classification and warrant further research.This study demonstrates that perfusion-related parameters (D* and f) are useful for indicating the severity of liver disease, and may have the potential to become a promising non-invasive tool for monitoring liver function.