Classification of Leukemia and Lymphoma using Faster R-CNN

Abstract

Blood disorders are disorders that impair the ability of the blood to function normally. There are various types such as leukemia and lymphoma, and the symptoms differ accordingly. Many death cases are related to these malignant tumors as a result of the difficult therapy and its negative effects. Furthermore, patients require a variety of diagnostic tests, which are costly, time-consuming, and can be stressful and hazardous to the patients. Furthermore, only a qualified pathologist can distinguish and identify various forms of lymphomas and leukemias. In this paper, the “Faster Region-Based Convolutional Neural Network; Faster R-CNN” algorithm is implemented to classify normal patients and ill patients into four categories: Acute Lymphoblastic Leukemia (ALL), Chronic Lymphocytic Leukemia (CLL), Follicular Lymphoma (FL), and Mantle Cell Lymphoma (MCL). This is done by dealing with Region of Interest (ROI) convolutional layer, then applying Vgg-16 model & Region Proposal Network (RPN) layer and adding a classifier layer. The dataset is based on microscopic and biopsies images. The proposed model achieved an accuracy of 87,575% after training in an acceptable computational time.