Visual attention-based leukocyte image segmentation using extreme learning machine

Abstract

The existence of computer aided automatic analysis came to light over a century ago and yet the acquisition of visual attention-based segmentation is still in its infancy despite mammoth advances in clinical leukocyte analysis, advanced neurocomputing and fast growing engineering technology. This paper presents a framework for novel segmentation model for leukocyte images using extreme learning machine ELM. The automatic analysis of microscopic leukocyte is an important task for diagnosticing many types of diseases namely leukaemia, malaria, psoriasis, AIDS, etc. The effective samples for training are obtained by saliency map of visual attention-based model. While in the learning stage, the effective samples are trained using ELM and segment leukocytes from blood smear images. The proposed framework is fully automatic and the experimental results outperform the state-of-the-art approaches.