Detection and Classification of Blood Cancer Using Deep Learning Framework

Leukocytes, which are generated in the bone marrow, account for around 1% of all blood cells. Blood cancer develops as a result of the uncontrolled multiplication of these white blood cells. The proposed study offers a reliable method to analyze the blood cancer. The two classes of blood cancer are acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML). Here a deep learning framework is employed to detect blood cancer automatically from input images and completely eliminate the possibility of human error. We have utilized the deep convolutional neural networks in our framework to distinguish leukemic cells from healthy blood cells. Then we have extracted the deep features from blood cancer images using CNN and trained the model followed by classification. We have conducted the experiments on publicly available dataset, namely SN-AM. Comprehensive experiments’ results demonstrate that our developed method has obtained the accuracy of 98.71%, recall of 97.48%, precision of 98.37%, and F1-score of 97.87%, achieving better performance compared to the previous state-of-the-art (SOTA) approaches.