Staging liver fibrosis with various diffusion-weighted magnetic resonance imaging models

Figure 1 Patient enrollment flow chart. DWI: Diffusion-weighted imaging; ROI: Region of interest; SF: Significant fibrosis; NSF: Non-significant fibrosis.

Figure 2 A 27-year-old female patient with hepatitis B virus for nine years. The liver fibrosis stage was diagnosed as F1. A: Pathology image, H&E-stained samples (original magnification × 100) of right lobe of liver, shows portal fibrosis; B: Mono-apparent diffusion coefficient map; C-E: Intravoxel incoherent motion (IVIM) model-derived true diffusion coefficient, IVIM model-derived pseudo-diffusion coefficient, and IVIM model-derived perfusion fraction maps; F and G: Diffusion kurtosis imaging (DKI)-derived apparent diffusivity and DKI-derived excess kurtosis maps; H and I: Stretched exponential model (SEM)-derived distributed diffusion coefficient and SEM-derived intravoxel heterogeneity index maps; J-L: Fractional order calculus model-derived diffusion coefficient, fractional order calculus (FROC)-derived fractional order parameter, and FROC model-derived microstructural quantity maps; M-O: Continuous-time random-walk (CTRW) model-derived anomalous diffusion coefficient, CTRW model-derived temporal diffusion heterogeneity index, and CTRW model-derived spatial diffusion heterogeneity index maps; P: The region of interest placement in the liver parenchyma.

Figure 3 The detailed comparisons of Mono-apparent diffusion coefficient, intravoxel incoherent motion model-derived true diffusion coefficient, fractional order calculus model-derived diffusion coefficient and continuous-time random-walk model-derived anomalous diffusion coefficient in different stages of liver fibrosis. A-D: Box-and-whisker plots showing the distributions of Mono-apparent diffusion coefficient (A), intravoxel incoherent motion model-derived true diffusion coefficient (B), fractional order calculus model-derived diffusion coefficient (C), and continuous-time random-walk model-derived anomalous diffusion coefficient (D) in various liver fibrosis stages (^aP < 0.05). ADC: Apparent diffusion coefficient; IVIM-D: Intravoxel incoherent motion model-derived true diffusion coefficient; FROC-D: Fractional order calculus model-derived diffusion coefficient; CTRW-D: Continuous-time random-walk model-derived anomalous diffusion coefficient.

Figure 4 The ability of the diffusion-weighted imaging -derived parameters and their combinations (diffusion-weighted imaging models) to detect significant fibrosis. Receiver operating characteristic curves for detecting significant fibrosis using diffusion-weighted imaging -derived parameters (A) and their combinations into full models (B). ADC: Apparent diffusion coefficient; IVIM-D: Intravoxel incoherent motion model-derived true diffusion coefficient; IVIM-D^*: Intravoxel incoherent motion model-derived pseudo-diffusion coefficient; IVIM-f: Intravoxel incoherent motion model-derived perfusion fraction; DKI-MD: Diffusion kurtosis imaging-derived apparent diffusivity; DKI-MK: Diffusion kurtosis imaging-derived excess kurtosis; SEM-DDC: Stretched exponential model-derived distributed diffusion coefficient; SEM-α: Stretched exponential model-derived intravoxel heterogeneity index; FROC-D: Fractional order calculus model-derived diffusion coefficient; FROC-β: Fractional order calculus model-derived fractional order parameter; FROC-μ: Fractional order calculus model-derived microstructural quantity; CTRW-D: Continuous-time random-walk model-derived anomalous diffusion coefficient; CTRW-α: Continuous-time random-walk model-derived temporal diffusion heterogeneity index; CTRW-β: Continuous-time random-walk model-derived spatial diffusion heterogeneity index; CTRW: Continuous-time random-walk; DKI: Diffusion kurtosis imaging; FROC: Fractional order calculus; IVIM: Intravoxel incoherent motion; SEM: Stretched exponential model.