Sickle Cell Classification Using Deep Learning

Artificial Intelligence has been a boon to healthcare for quite a long time. While AI has the potential to assist in several domains, blood smear analysis has several challenges that need to be addressed to ensure the accuracy, interpretability, safety, and ethical use of AI in this context. Proper validation, overlapping of cells, data availability, addressing biases, interpretability, regulatory compliance, workflow integration, and ethical considerations are important aspects that must be carefully considered when using AI in blood smear analysis. One of the lesser-tackled problems through Artificial Intelligence is the classification of sickle cells. Sickle cell disease is a genetic disorder affecting the hemoglobin resulting in a reduced supply of oxygen to the entire body. Currently, there is no cure for sickle cell disease, and treatment is focused on managing symptoms and preventing complications. Manual identification and classification of sickle cells in blood smear images can be time-consuming and prone to human error. Hence, there is a need for automated methods to classify sickle cells and tackle these problems. This paper discusses a deep learning model to detect the presence of sickle cells in the blood, thereby classifying them. The model focused on here is ResNet50 giving a test accuracy of 93.88%.