Analysis of Blood Samples to Identify Leukemia Cells using a Support Vector Machine and K-Nearest Neighbor Algorithms

Abstract

The study proposes an innovative approach using MATLAB to automate the counting of leukemia cells in blood samples, employing Support Vector Machine (SVM) and Nearest Neighbor algorithms. The method involves preprocessing blood sample images to enhance contrast and apply image filters, followed by segmentation techniques for isolating individual cells. SVM and nearest neighbor algorithms are trained using extracted features such as cell size, shape, and texture. Accurate detection and counting of leukemia cells play a crucial role in leukemia diagnosis and management. Leukemia is a group of cancers characterized by abnormal white blood cell proliferation in the bone marrow, leading to symptoms like bleeding, bruising, fatigue, and increased infection risk due to insufficient normal blood cells. Diagnosis typically involves blood tests or bone marrow biopsy. In clinical bioinformatics, SVM algorithms have enabled the development of robust experimental cancer diagnostic models, utilizing gene expression data with a small number of samples and numerous variables. Efficient implementations of SVM algorithms further facilitate practical application. Support Vector Machines excel in mapping data to higher-dimensional spaces through kernel functions, allowing the identification of maximum-margin hyperplanes for separating training instances.