An Effective Approach for Detecting Acute Lymphoblastic Leukemia Using Deep Convolutional Neural Networks

Abstract

In this paper, we present a fully automatic acute lymphoblastic leukemia detection method based on deep neural networks. ALL is a condition affecting the leukocytes. Children tend to be prone to have this melanoma. The fundamental problem with this type of cancer is that, unlike other cancers, it does not create tumors, making it extremely difficult to identify. Prior to automation, manual microscopic testing procedures were used, but they were time-consuming and error-prone. To overcome this issue, many automated systems were introduced, which used machine learning techniques. But, because we are dealing with medical information, we may require better efficiency and accuracy, so as an improvement, our proposed system employed several convolutional neural networks architectures. In the proposed system, five such CNN algorithms were implemented to classify and separate the cancerous and non-cancerous cells. The system accepts the blood cell image from the user and predicts whether the cell contains one/more blasts depending upon the prediction value obtained from the CNN algorithm on the stained cell image. As a result, it significantly reduces the research costs, increases the speed of testing, and can be a lifesaver for millions of cancer patients.