Automatic Color Nuclei Segmentation of Leukocytes for Acute Leukemia

In this study, we used an efficient and simple technique for the automatic segmentation of nuclei of Leukocytes, which is not only accurate in the results but also fast as compared to other published algorithms. Leukemia is a blood cancer and has several types but all types begin from the cells in the bone marrow. Hematology, the study of blood is an important step for the diagnosis of various diseases especially for Leukemia. Manual Hematology is done by experts for any blood disorder, but is time consuming, high mental and physical labor is required, highly subjective and erroneous. The diagnosis of any diseases required high accuracy but Leukemia is a life critical disease and has zero tolerance for the errors. Automatic diagnosis of microscopic images through digital image processing techniques is the need of the day. Segmentation is one of the most important and challenging technique in the automatic diagnosis of Leukemia. The developed method is tested on 380 images and the accuracy of various types of Leukocytes (WBCs) was found out qualitatively. Also the developed technique is quantitatively evolved for speed performance. The segmentation accuracy of the developed technique is 96.5% while the efficiency reduces the computation time by 50% as compared to other published techniques. The algorithm along with the data set is published on MATLAB file exchange for an evaluation.