Detection of leukaemia and its types using image processing and machine learning



Introduction This review focuses on how image processing and machine learning can be useful for the morphological characterization and automatic recognition of cell images captured from peripheral blood smears. Methods The basics of the 3 core elements (segmentation, quantitative features, and classification) are outlined, and recent literature is discussed. Although red blood cells are a significant part of this context, this study focuses on malignant lymphoid cells and blast cells. Results There is no doubt that these technologies may help the cytologist to perform efficient, objective, and fast morphological analysis of blood cells. They may also help in the interpretation of some morphological features and may serve as learning and survey tools. Conclusion Although research is still needed, it is important to define screening strategies to exploit the potential of image-based automatic recognition systems integrated in the daily routine of laboratories along with other analysis methodologies.

Authors: Rodellar, J., Alférez, S., Acevedo, A., Molina, A., Merino, A.