Segmentation of tuberculosis bacilli in Ziehl-Neelsen tissue slide images using Hibrid Multilayered Perceptron network

Abstract

Segmentation of Zeihl-Neelsen tissue slide images is an important step in computer-assisted tuberculosis bacilli detection. In Zeihl-Neelsen tissue slide image, colour is the most prominent feature to detect the presence of tuberculosis bacilli from its image. In this paper, an automatic colour image segmentation using neural network is proposed. The segmentation is based on supervised approach using Hybrid Multilayered Perceptron network. Colour images are converted to HSI colour space before being segmented by the network. The colour image are characterised in HSI colour space for better colour representation. Experimental results indicate that the proposed method provides good segmentation on variety of Ziehl-Neelsen tissue slide images.