Design and Optimization of Hybrid Deep Convolutional Neural Networks for Efficient Leukemia Multiclass Classification

Leukemia is a cancer of the bone marrow that leads to the production of a large number of abnormal white blood cells interfering with the normal blood cell production in the bone marrow. The disease is diagnosed by the doctor on the basis of clinical evaluation, Complete Blood Count test and bone marrow biopsy. Leukemia majorly classifies into two types based on the kind of white blood cells namely Acute Lymphoblastic Leukemia and Acute Myeloid Leukemia. The authors propose to utilize pretrained networks like Inception V3 and ResNet152 for image processing and classification to classify ALL subtypes such as L1, L2 and L3 and AML subtypes such as M2, M3 and M5 . The intention of the authors is to design and develop an efficient deep learning framework for classifying leukemia with reduced data from American society of hematology with improved accuracy.