Detection and Spatial Analysis of Hepatic Steatosis in Histopathology Images using Sparse Linear Models

Abstract : Hepatic steatosis is a defining feature of nonalco-holic fatty liver disease, emerging with the increasing incidence of obesity and metabolic syndrome. The research in image-based analysis of hepatic steatosis mostly focuses on the quantification of fat in biopsy images. This work furthers the image-based analysis of hepatic steatosis by exploring the spatial characteristics of fat globules in whole slide biopsy images after performing fat detection. An algorithm based on morphological filtering and sparse linear models is presented for fat detection. Then the spatial properties of detected fat globules in relation to the hepatic anatomical structures of central veins and portal tracts are explored. The test dataset consists of 38 high resolution images from 21 patients. The experimental results provide an insight into the size distributions of fat globules and their location with respect to the anatomical structures.