Evaluation of Liver Fibrosis Using Texture Analysis on Combined-Contrast-Enhanced Magnetic Resonance Images at 3.0T

Figure 8

Illustrative examples of texture feature classes for fibrosis prediction. A 54 year-old female with cirrhosis. (c) Standardized regions of interest (ROI) images, without transformation (raw), with gradient and Laplacian transformations. (b) Corresponding pixel intensity histogram and its Gaussian mixture model fit with two normal populations. (a) Voronoi polygons constructed on the corresponding ROI images. These texture classes were found to be predictive of liver fibrosis on CCE images (see text). For each texture class (intensity histogram, Gaussian mixture, and Voronoi polygons), relevant statistics were calculated as detailed in supplementary materials and were used for fibrosis prediction.