Using Convolutional Neural Networks for Automatic White Blood Cell Cancer Detection in Bone Marrow Micrographs

Abstract

The bone marrow is responsible for producing about 1% of all blood cells, which are known as leukocytes. Blood malignancy originates in the unchecked multiplication of these neutrophils. The suggested research offers a reliable method for categorizing three distinct kinds of cancer. Multiple Myeloma (MM) and Acute Lymphoblastic Leukemia (ALL) use the SN-AM dataset. In cases of acute lymphoblastic leukemia (ALL), an abnormally high number of lymphocytes are produced by the bone marrow. However, unlike other cancers, multiple myeloma (MM) causes malignant cells to collect in the bone marrow rather than being disseminated throughout the body. As a result, they inhibit the development of new, healthy blood cells by crowding them out. In the past, this would take a very long time and a lot of effort on the part of a trained expert. Using deep learning methods, specifically convolutional neural networks, the suggested model eliminates the possibility of mistakes in the human process. The algorithm is educated on pictures of cells, so it knows how to best pre-process and retrieve characteristics from those images. Next, the sort of malignancy in the cells is predicted by training the model using an improved version of the Dense Convolutional neural network (DCNN) architecture. While remembering the samples precisely 94 times out of 100, the model accurately reproduced all the data.