Recent Artificial Intelligence Advances in Detection and Diagnosis of Sickle Cell Disease: A review

Sickle cell anemia is a genetic disease characterized by a genuine alteration of hemoglobin that leads to the emergence of sickle-shaped red blood cells. The clinical diagnosis of sickle cell disease is based on blood analysis tests such as electrophoresis, chromatography and isoelectric focusing, blood count scanning or peripheral blood smears. These analyses aim to identify the presence of the hemoglobin S (HbS) in the sickle cell patient's blood. Nonetheless, these approaches face limitations linked to processing delays, substantial expenses, the need for specialized expertise, and inaccessibility in developing countries, where the prevalence of sickle cell disease is substantial. Thus, artificial intelligence techniques, including medical image segmentation and machine learning algorithms, offer innovative solutions to address these challenges. This paper gives an overview of recent advances related to image segmentation, feature extraction as well as classification approaches used in the detection and diagnosis of sickle cell disease. We introduced both computer vision techniques, machine learning and convolutional neural networks for sickle cell detection in microscopic images of blood smear. We discuss specific challenges related to segmentation methods while processing microscopic images containing overlapping cells. In addition, we examined challenges related to the robustness of convolutional neural network models used for feature extraction and then classification of sickle cells images. To conclude this paper, research prospects that can be explored in the future related to sickle cell detection in medical concerns are given.