Enhanced Support Vector Machine Based Leukaemia Cancer Classification

Abstract

Leukaemia is a blood cancer that is characterized by the bone marrow’s unregulated and irregular generation of white blood cells (leukocytes) in the blood. By testing the microscopic blood cell images, diseases can be detected and its early diagnosis can be done. In the current work,at smart decision support systemutilizing microscopic images is proposedto make a diagnosisof acute lymphoblastic leukaemia. Identification through images is a quick and inexpensive technique since there is no equipment is specially needed for lab testing. The improved Bare-Bones Particle Swarm Optimization algorithm is usedto identify the important distinguishing featuresof blast cells and the healthy cells to help in carry out the acute lymphoblastic leukaemia classification with efficiency. To avoid the poor accuracy, the modified median filtering has been introduced in this novel, which helps in removing the noise from the images and maintains the edge which is used to improve the identification of leukocytes and lymphocytes during the segmentation process. Enhanced linear contrast stretching is introduced in image enhancement for enriching the image. In the next step, feature extraction is carried out through the second order statistical features. Bare-bones with Adaptive Bat Optimization (BBABO) is used for feature selection. At last, the classification is carried out by using the combined Fuzzy Neural Network (FNN) and here the Enhanced Support Vector Machine (ESVM) is a classifier called as FNESVM.