Computer Aided System for Leukocytes Classification and Segmentation in Blood Smear Images

Abstract

Detection and counting of white blood cells (WBC) in blood samples provides valuable information to medical specialists, helping them to evaluate a wide range of important hematic pathologies such as AIDS and blood cancer (Leukaemia). However, this task is prone to errors and time consuming. An automatic detection and classification of WBC images can enhance the accuracy and speed up the detection of WBCs. In this paper, we propose an efficient framework for localization of WBCs within microscopic blood smear images using a multi-class ensemble classification mechanism. In the proposed framework, the nuclei are first segmented, followed by extraction of features such as texture, statistical, and wavelet features. Finally, the detected WBCs are classified into five classes including basophil, eosinophil, neutrophil, lymphocyte, and monocyte. Experimental results on a natural (non-synthetic) benchmark database validate the effectiveness and efficiency of the proposed system in contrast to state-of-the-art schemes.