Abstract
Leukemia is a type of blood cancer that affects the bone marrow, lymphatic system, and red blood cells. The medical field is progressing very fast in medical imaging technology, even after the progress there is a need for more accurate and faster methods for leukemia classification and detection. Early and accurate detection of leukemia is crucial for timely intervention and to reduce the mortality rate of the affected patients. Blood tests are essential for diagnosis of Leukemia, where a drop of blood is smeared on slide and looked under the microscope by a lab technician. The traditional methods of detection are dependent on manual interpretation of slides which is time consuming and error prone at times based on expertise level.
This project aims at achieving highest accuracy to detect Leukemia with almost negligible error rates. In this project, an application is developed for the medical authorities, where the blood sample of a patient can be uploaded, and the type of Leukemia can be confirmed. This technology uses some deep learning algorithms such as Vision Transformers(ViT), Convolutional Neural Networks(CNN), ResNet and VGGNet for Leukemia detection. The dataset is sourced from an open-source website working on collecting huge amounts of microscopic images in the medical domain. They are devoted to collect valid medical data from various medical institutions from Iran. They have more than 40,000 microscopic images. The dataset has a variety of images for Leukemia classification into four categories namely acute lymphoblastic leukemia, acute myeloblastic leukemia, chronic lymphocytic leukemia, and chronic myelogenous leukemia.