Ciliated epithelium segmentation using texture features classification

Abstract

This article deals with the design of automated methods for the segmentation of ciliated epithelium in microscopic images. The proposed method is based on texture analysis of image and subsequent classification of these parameters into 3 classes: epithelial cells, cilia and background. The aim of the work is to find the best combination of initial conditions of segmentation. An output of the algorithm is an original microscopic image with highlighted regions assumed to be cilia.