Detecting Acute Lymphoblastic Leukemia Through Microscopic Blood Images Using CNN

Abstract

Leukemia refers to cancer, takes place when bone marrow generates excess amount of abnormal blood cells. These abnormal blood cells grow rapidly interrupting the functioning of the other normal cells. Hence, body loses its capability to fight with exterior organism like viruses, bacteria, fungi, and so on, hereby, affecting the immunity of the person. Leukemia, if not treated on time can lead to death. This disease affects adults as well as children. Thus, early detection of the disease would help in proper treatment. The manual diagnosis of this disease is quite laborious, tedious, and time-consuming which involves blood test and biopsy. Thus, to overcome these shortcomings, this paper discusses a computer-aided automated diagnosis system for detection of acute lymphoblastic leukemia (ALL) using deep-learning models. A pretrained AlexNet model is deployed for performing this task. Experiment is done using microscopic blood cell images. This overall framework helps in detecting the malignant cells easily. From the experimental results, it is evident that this proposed method achieves an accuracy of 98% without the use of any image segmentation technique or feature extraction technique. Hence, the work reported in the paper would provide a framework to aid pathologist in diagnosing acute lymphoblastic leukemia accurately and quickly.