The field of biomedical diagnosis has become very important in the evolution of medicine, and especially for the detection of cancer cells. In this paper, we will develop an algorithmic treatment by segmentation, using elements of mathematical morphology. The aim of our work is to obtain a maximum rate of recognition of leukemia. Our algorithmic diagram can segment, detect, characterize and describe cancerous blood cells (leukemia). This proposal is an important task in the interpretation and diagnosis of pathologies for hematologists. The experimental results applied on our microscopic medical images are encouraging for the identification of abnormal blood cells generally, and the distinction of leukemia especially, which shows the efficiency and rapidity of our algorithmic system.