Blood cancer also known as leukaemia, involves excessively high levels of aberrant white blood cells, which disrupts the blood’s normal functioning and causes damage to the bone pith. The excessive production of aberrant and immature white blood cells harms the immunological system by reducing the bone marrow’s capability to produce red blood cells and platelets. Usually, manual techniques like complete blood counts, marrow aspiration of bone, or microscopic analysis of blood smears are used to diagnose haematological malignancy. While these manual diagnostic methods are cost-effective, they are often considered Less trustworthy, time-intensive, and labour-intensive. Advancements in medical technology have successfully resolved these challenges in the past. The limitations of manual blood smear diagnosis for detecting blood cancer have been addressed through the development of automated methods leveraging machine learning techniques for more effective and trustworthy leukaemia diagnosis. Over the past years, various methods have been proposed for automated systems aimed at enhancing preprocessing, segmentation, attribute extraction, attribute selection, and improving classification accuracy in blood cancer detection. The categorization of white blood cells is thoroughly reviewed in this work, which also explores several machine learning and deep learning approaches. The review examines these methodologies based on their effectiveness in pre-processing, segmentation, attribute extraction and attribute selection, and the overall performance of automated systems.