Classification of Effective Cells in Diagnosis of Chronic Myeloid Leukemia (CML) Using Semi-automatic Image Processing of Microscopic Images

• Published in: مجله دانشکده پزشکی اصفهان
• Main Authors: Narjes Ghane, Alireza Vard, Ardeshir Talebi, Pardis Nematollahy
• Format: Article
• Language: Persian
• Published: Isfahan University of Medical Sciences 2017-01-01
• Subjects: Chronic myeloid leukemia; Classification; Granulocytic category; Image processing
• Online Access: http://jims.mui.ac.ir/index.php/jims/article/view/6978

Background: Chronic myelogenous leukemia (CML) is a type of blood cancers that usually occur in adults. The first step for diagnosis of leukemia is blood test and counting cells. Diagnosis and counting cells from blood smear are done by pathologist using optical microscope. This is a time-consuming and costly process and needs experience and expert in this field. Besides, other factors such as fatigue and working conditions can negatively affect the diagnostic evaluation. Thus, an aid tool for pathologists to help in the diagnostic process can be so useful. This research proposed a novel software tool to diagnose and classify of chronic myeloid leukemia cells. Methods: In the proposed method, after accurate manual segmentation, various geometric features of cell and nucleus were extracted from neutrophils cells using image processing techniques. Then, applying these features by a new designed tree classifier, cells were categorized in to six groups. Findings: The proposed method was evaluated on 120 blood smear microscopic images including 714 white blood cells (WBCs). An accuracy of over 97%, specificity of over 98% and sensitivity of over 91% for all of six groups were achieved. Conclusion: In this study, a semi-automatic method was proposed for detection and classification of effective cells in diagnosis of chronic myeloid leukemia in microscopic images utilizing image processing methods.