SSRG International Journal of Electronics and Communication Engineering ( SSRG – IJECE ) – Special Issue ICCREST Mar 2019 ISSN: 2348 – 8549 www.internationaljournalssrg.org Page 7 Detection of Acute Lymphocytic Leukemia using Statistical Features S. Hariprasath #1 , T. Dharani *2 , Dr. M. Santhi #3 #1 Assistant Professor, PG Scholar *2 , Professor: HOD of ECE #3 & Dept. of ECE Saranathan College of Engineering, Trichy, India Abstract Accurate diagnosis of Leukemia is an important issue in the medical field in order to provide effective treatment to the patient. Leukemia is caused due to the abnormalities in the lymphatic (immune) system in our body. The chance of getting affected by leukemia is more common in children than adults. Cell types that involved in leukemia are white blood cells which are potent infection fighters. When leukemia causes abnormal production of WBCs which do not function properly is known as Acute Lymphocytic Leukemia (ALL) whereas, in other type of leukemia called Chronic Lymphocytic Leukemia (CLL), immature WBCs are capable of performing their functions normally. When compared to Chronic Leukemias, Acute Leukemias are more hazardous. In this paper Acute Lymphocytic Leukemia (ALL) is focused. In this work, from the given dataset that consisting of both benign (healthy) & malignant cells, Leukemic cells are detected & classified based on blast cells’ morphology. There are various image processing techniques to detect leukemia and its types. Linear, SVM and classifiers are analyzed. SVM-R produced accuracy, sensitivity and specificity of 86.67%, 85%, and 90% respectively for noisy data. Keywords - Leukemia, Pattern Classification system, linear classifier, SVM. I. INTRODUCTION White Blood Cells has the major contribution to the immune system of our body. It also helps in detecting diseases such as leukemia. The death rate in our country has been increasing in every year due to many hazardous diseases. Leukemia (Blood Cancer) is considered as the most threating disease since it may cause immediate death. It causes the production of abnormal white blood cells that are responsible for fighting diseases. Leukemic cells continue to grow, multiply & divide and results in decreasing the capability to fight against infections, control bleeding and transport oxygen [1]. Different types of leukemia depend on the type of blood cell that results in cancer. This classification is based on two systems namely: French-American-British (FAB) classification and World Health Organization (WHO) [2].When leukemia occurs in lymphocytes, it is known as lymphocytic leukemia. Hence further subtypes of lymphocytic leukemia are Acute Lymphocytic Leukemia (ALL) and Chronic Lymphocytic Leukemia (CLL). The other types of leukemia are known as Acute Myeloid Leukemia (AML) and Chronic Myeloid Leukemia (CML) that occurs due to abnormal myelocytes [3]. According to FAB classification, the subtypes of ALL & AML are L1, L2 and L3, and M1, M2, M3, M4, M5, M6 and M7 respectively. Since Acute Leukemias are more hazardous, it is necessary to diagnose them at the early stage for medication [10]. By observing the blood cells from microscopic images, many diseases can be diagnosed and treated. In this work, only ALL is considered which affects cells called lymphocytes. ALL primarily affects children and adults over 50 years of age. ALL can affect children who are younger than 5 years of age and declines and begins to rise again after 50 years of age. Hence ALL can even leads to death when it is untreated. Early diagnosis of ALL is very crucial for a person‟s recovery. This diagnosis and detection of ALL is based on identification of lymphoblasts by microscopy. The best way to distinguish malignant cells from benign cells is to utilize the morphological features of a blast cell nucleus where immature white blood cells are known as „blast cells‟ [4]. The discrimination between healthy and malignant cells can be made by using their cell structure. The differences in the cells structure of healthy and malignant cell is shown in figure 1. Figure 1. Difference between healthy and malignant cell structure II. EXISTING SYSTEMS Various morphological, shape and textural features have been taken into account for the detection and classification of a disease. Commonly used cancer database for detection and classification is ALL-IDB. In „ALL- IDB: the acute lymphoblastic leukemia image database for image processing‟ by Ruggero donida