Multifractal feature descriptor for grading Hepatocellular carcinoma

This paper presents a textural feature descriptor that can be effectively utilized for grading Hepatocellular carcinoma (HCC) histopathological images. The proposed feature descriptor observes the local and spatial characteristics of the texture by utilizing multifractal computation, and it is incorporated with a bag-of-feature (BOF)-based classification model to classify a set of images. We compare the proposed feature descriptor with four well-founded feature descriptors in the experiments, and benchmark the classification performances. The benchmarked results indicated the significance of the multifractal feature descriptor.