RECENT DEVELOPMENTS IN THE DETECTION AND CLASSIFICATION OF LEUKEMIA - A COMPREHENSIVE EVALUATION OF DEEP LEARNING SOLUTIONS

Abstract

The blood is essential for keeping us alive since it carries oxygen, nutrients, and immune cells around the body. Leukemia is a type of cancer that affects the tissues that make blood. It can be deadly if not caught early. So, finding it early is for getting the right therapy and living longer. Blood smear analysis using a microscope has long been a major way to diagnose diseases. However, looking at these pictures by hand takes a lot of time, is easy to make mistakes, and isn't always consistent. Machine learning (ML) and other automated methods have been used to solve these problems by looking at enormous datasets and finding patterns that are difficult to see by hand. But ML methods frequently need people to choose the right features, and they might not work well with data that is complicated and has a lot of dimensions. Deep learning (DL) solves these problems by automatically learning important aspects from raw medical images. This makes it easier and faster to find leukemia. DL models, notably convolutional neural networks (CNNs), have shown that they can classify different types of leukemia better than other methods with little help from people. This article gives a full picture of the most recent improvements in DL methods for finding leukemia and shows how they could change the way blood tests are done. Also, a comparison of several DL models is shown, along with their pros and cons, to that they can still provide more accurate diagnoses and lower the chance of human error. There is also a discussion of the problems and opportunities in this sector, with a focus on how DL can help find leukemia early and accurately.