SELECTION THE BEST FEATURES FOR LEUKOCYTES CLASSIFICATION IN BLOOD SMEAR MICROSCOPIC IMAGES Omid Sarrafzadeh* a , Hossein Rabbani a , Ardeshir Talebi b , Hossein Yousefi-Banaem a a Dept. of Biomedical Engineering, Faculty of Advanced Medical Technology, Isfahan University of Medical Sciences, Isfahan, Iran; b Dept. of Pathology, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran ABSTRACT Automatic differential counting of leukocytes provides invaluable information to pathologist for diagnosis and treatment of many diseases. The main objective of this paper is to detect leukocytes from a blood smear microscopic image and classify them into their types: Neutrophil, Eosinophil, Basophil, Lymphocyte and Monocyte using features that pathologists consider to differentiate leukocytes. Features contain color, geometric and texture features. Colors of nucleus and cytoplasm vary among the leukocytes. Lymphocytes have single, large, round or oval and Monocytes have singular convoluted shape nucleus. Nucleus of Eosinophils is divided into 2 segments and nucleus of Neutrophils into 2 to 5 segments. Lymphocytes often have no granules, Monocytes have tiny granules, Neutrophils have fine granules and Eosinophils have large granules in cytoplasm. Six color features is extracted from both nucleus and cytoplasm, 6 geometric features only from nucleus and 6 statistical features and 7 moment invariants features only from cytoplasm of leukocytes. These features are fed to support vector machine (SVM) classifiers with one to one architecture. The results obtained by applying the proposed method on blood smear microscopic image of 10 patients including 149 white blood cells (WBCs) indicate that correct rate for all classifiers are above 93% which is in a higher level in comparison with previous literatures. Keywords: Blood smear microscopic image, feature extraction, fuzzy C-means clustering, leukocytes (WBCs) segmentation and counting, SVM classifier. 1. INTRODUCTION Blood consists of three types of cells and cell fragments floating in a liquid called plasma. These cellular components are: Red Blood Cells ("erythrocytes", "RBCs"), White Blood Cells ("leukocytes", "WBCs") and Platelets. WBCs play a significant role in the diagnosis of different diseases such as leukemia and different types of infections [1], so extracting information from them is valuable for hematologists. WBCs composition also reveals important diagnostic information about patients. Substituting automatically detecting and counting of WBCs for manually locating and counting different classes of WBCs is an important topic in the domain of cancer diagnosis [2]. Microscopic differential WBC count is still performed by hematologists, being indispensable in diagnostics with malignance suspicious. This as a reference method is slow and subjective and its reproducibility is poor, however its value for blood samples containing abnormal cells remains indisputable. Therefore, automation of this task is very helpful for improving the hematological procedure and accelerating diagnosis of many diseases [3]. WBCs contain nucleus and cytoplasm and there are five types of leukocytes (WBCs) found in the blood: Neutrophils, Basophils, Eosinophils, Lymphocytes, and Monocytes. Each cell type has a specific role to play in our body's immune system [4]. The texture, color, size and morphology of nucleus and cytoplasm make differences among WBCs. The segmentation step is very crucial because the accuracy of the subsequent feature extraction and classification depends on the correct segmentation of WBCs. It is also a difficult and challenging problem due to the complex nature of the cells and uncertainty in the microscopic images. Therefore, this step is the most important challenge in many works in the literatures and improvement of cell segmentation has been the most common effort in many researches. Many blood smear image segmentation methods have been proposed in the area of general segmentation of WBCs which are generally based on edge and border detection, region growing, filtering, mathematical morphology, and watershed clustering [5-14]. Despite many beneficial explorations have been carried out in WBCs segmentations, majority of them have some defects such as complexity of arithmetic, difficulty to ensure parameters and