Leukemia Detection Algorithms: A Comparative Study

Leukemia is a group of blood cancers that usually begins in the bone marrow and is caused by a high number of abnormal blood cells. Yearly, the number of new cases of Leukemia increases with a high mortality rate due to the late detection of the disease. However, computerizing disease detection decreases the number of mortal cases by implementing several methodologies and algorithms. Image processing techniques are one of the major disciplines in detecting the disease early. A set of related algorithms were developed to help detect Leukemia depending on several blood components features like cells' properties (shape, color, size), cytoplasm features, and others. This work compares two approaches to Leukemia detection: Color K-means Clustering (CKC) and Acute Lymphoblastic Leukemia Subtypes (ALLS) detection algorithms. CKC algorithm depends on morphological filtering and segmentation using color k-means clustering and is tested with Nearest Neighbor (KNN) classifier. ALLS is based on detecting Leukemia subtypes by going through several image processing stages that lead to classifying the image as infected or not. In order to test the amount of accuracy difference between the two approaches, the extracted features of the testing set of images are processed using Weka software. The precision, recall, f-score, accuracy, and time measures were computed for comparison purposes.