LEUKEMIA DISEASE CLASSIFICATION AND DETECTION IN BLOOD MICROSCOPIC IMAGES USING DEEP LEARNING TECHNIQUES

Abstract

The early identification and diagnosis of leukaemia is a key challenge in the field of sickness diagnostics. This challenge entails making an accurate differentiation between healthy and cancerous leukocytes in the early stages of the disease while keeping expenses to a minimum. In spite of the widespread occurrence of leukaemia, there are only a limited number of flow cytometers, and the diagnostic procedures followed in laboratories are time-consuming. Cancer of the bloodforming tissues, sometimes known as leukaemia, can affect the lymphatic system as well as the bone marrow. If it is discovered at an earlier stage, treatment may be more successful. As a result of this research, a new classification model for blood microscopic images was established. This model makes a distinction between images unaffected by leukaemia and those that are afflicted by leukaemia. The detection times and accuracy of machine learning and deep learning methods are far higher than those of human analysts. In order to address this scenario, we have used deep learning based Leukemia disease classification and detection in blood microscopic images. This is by comparing two models, namely traditional CNN and deep CNN model (Alex Net and VGG-16 Net model). According to the C-NMC 2019 dataset, a total of 11,154 blood microscopic images were collected for the purpose of evaluating our proposed approach. Based on our study findings, it has been noted that the performance of a VGG-16 Net model, surpasses that of other two models such as the Traditional CNN and Alex Net model. The optimal performance of the model is achieved by the utilization of VGG-16 Net model as a feature extractor, and Soft-max as the classifier. This configuration yields an accuracy of 97.44, precision of 97.5, recall of 97.5, and F1-score of 97.5