An aplication of self-similarity analysis for segmentation of images of human liver

Abstract

Images of human liver taken from computed tomography are considered. These images are analysed as self-similar sets. To accomplish this analysis, the multifractal mathematical model is applied. The main task behind this article is investigation of relevance of multifractal mathematical model for discrimination of different tissues of human liver. Indeed, we need to show that a multifractal spectrum allows us to distinct different types of tissues. This allows us to suggest application of multifractal model for image segmentation task. Using multifractal spectra was shown that three different types of tissues can be identified. Types of tissues under investigation were: liver, tumour, and surrounding tissues i.e. abdomen.