Unlocking the Potential: Machine Learning and Deep Learning in Leukemia Diagnosis with Explainable AI

The human body can undergo changes in blood cell structure and characteristics when affected by contaminants. Examining the tiny images of blood cells helps identify potential infections or irregularities within the body, aiming to detect diseases. Accurately segmenting these cells significantly enhances disease detection, making it more precise and robust. Microscopic analysis of blood cells plays a pivotal role in pathological examinations, facilitating the identification of specific diseases and abnormalities, crucial for diagnosing various disorders, planning treatments, and assessing their outcomes. Leukemia is a disease that produces the infections within the bone marrow of a human being. Bone marrow generates the leukocytes (White Blood Cells), erythrocytes (Red Blood Cells), and other components of blood. Due to leukemia infection, this blood-forming capacity of bone marrow is hampered. This chapter reviews the diverse applications of microscopic blood cell imaging in disease detection. It provides a brief overview of blood composition, followed by a generalized approach to analyzing microscopic blood cell images for specific medical imaging purposes. Leukemia systematics, symptoms, and different types are explored in the survey. Additionally, the chapter discusses and compares different methodologies proposed by researchers for disease detection through the analysis of microscopic blood cell images. Different machine learning, deep learning methods are explored in relation to leukemia diagnosis. Prominent phases of leukemia diagnosis, including segmentation and classification, are reviewed, and different algorithms employed for these phases are explored. The generalized methodology of disease detection along with a survey of state-of-the-art methods. As the explanability and interpretability issues make the deep learning frameworks as black boxes, revealing these frameworks’ black box nature is necessary to instill trust in the diagnosis decisions. Various explainable AI frameworks are explored in this chapter.