International Journal of Advance Research In Science And Engineering http://www.ijarse.com IJARSE, Vol. No.4, Special Issue (01), March 2015 ISSN-2319-8354(E) 1231 | Page A COMPARATIVE STUDY OF TECHNIQUES FOR LEUKAEMIA DETECTION Vasundhara Sharma 1 , Sonu Elsa Jacob 2 and Praveen Kumar Sharma 3 1,2 Student, Department of Electronics and Communication Engineering, 3 Assistant Professor, Department of Electronics and Communication Engineering, B.K. Birla Institute of Engineering and Technology, Pilani, Rajasthan, (India) ABSTRACT Leukaemia is a cancer of blood that causes more deaths than any other cancer. Presently, the diagnosis of blood samples is done through visual examination by doctors. For proper and efficient treatment of leukaemia it is essential to detect it in early stage and proceed with the monitoring and evaluation. Till date many research have been done to design an automated system for detection of leukaemia either through study of microscopic images of blood samples or bone marrow biopsy. In this paper, a literature review has been done to classify the types of leukaemia and the various methodologies being followed by the researchers to detect it. Also, we have discussed some of the issues faced by the researchers. This paper also describes the various changes on texture, geometry, color and statistical analysis of microscopic images. Keywords: ALL, AML, Microscopic Imaging, MLP, SFAM. I INTRODUCTION Over the past few decades there has been huge growth in the application of image processing techniques in the bio-medical area. Researches are being carried out to develop efficient and cost effective system for solving medical problems. Presently the diagnosis of the blood samples is done through visual examination by the haematologist. However these morphological or biochemical analysis are subjected to various short comings like operators experience, tiredness, fatigue and slowness and these cell images are prone to have errors due to lack of efficiency, difficulties in cell nature and problems related to preparation of staining of blood cell slides [1][2][3]. Also, it has been found that the manual recognition method has an error rate between 30% and 40%. [4].This situation is further gearing up the demand of developing an automated system to provide more accuracy and precision. One of the most feared disease by the human is cancer. Leukaemia is the cancer of blood, and if not detected in the early stage can even lead to death. Automated systems based on artificial vision can speed up this operation [5]. Most of the symptoms and conditions of a disease are reflected in the blood. Studies have shown that all the techniques developed for medical imaging uses all the information about blood for classifying various diseases like leukaemia, anaemia, cancer, thalassaemia etc. These parameters may be RBC count, haemoglobin level or