Image Segmentation for Acute Leukemia Cells Using Color Thresholding and Median Filter

Abstract

Acute leukemia is a kind of the malignant disease which may lead to death due to its characteristic of rapid development of immature blood cells. Recently, several image processing techniques have been implemented to assist the task of acute leukemia diagnosis. The segmentation of acute leukemia cells is an important key to determine the accuracy of its classification task. This paper proposed a combined technique of color thresholding based on the RGB color information from acute leukemia slide images and median filter to segment the leukemia cells from the unwanted regions such as background and red blood cells. The presented results proved that the proposed technique was successfully segmented the acute leukemia cells from the Acute Myeloid Leukemia and Acute Lymphocytic Leukemia slide images, with the average accuracy rate of 97.63% and 97.64% respectively. Therefore, the proposed image segmentation technique could benefits the classification process of acute leukemia.