A Novel Framework for Sickle Cell Identification and Segmentation Using Deep Learning Techniques in Red Blood Cell

Abstract

Sickle Cell Anemia(SCA) is a genetic blood disorder problem as it changes hemoglobin molecule in red blood cell which leading sickle cell. Sickle cell genes affect the vital chemical which is necessary for blood it called hemoglobin. It reduces the creation of hemoglobin which is placed in red blood cells. Person having sickle cell genes force the human body to produce abnormal hemoglobin known as HbS (normal hemoglobin is known as HbA). HbS acts in a different way from HbA. As haracteristics of HbS make the red blood cells into change shape in its place of the normal shape. It converted into sickle shaped and also recognized as crescent moon. newlineDifferences in cell morphology between healthy and pathological cells make it possible to perform image-based diagnosis, which is very important for faster and more accurate diagnosis of potential SCA. Image-based analysis of SCA is capable of performing highly specific and sensitive sickle and normal erythrocyte classification through cell shape statistics on the stains and artifacts spreading on the microscopic images. It is to address the challenge of improving the prediction model to detect the sickle anaemia disease and providing timely response in predicting the disease. newlineInitially Sickle Cell Detection using Yolov5 technique is employed to detect the sickle anaemia on analysis of the cell morphology. Cell morphology has a key role in the detection of the sickle cell because the shapes of the normal blood cell and sickle cell differ significantly. It categorizes erythrocytes into two groups: round (normal) and elongated (sickle cells) on generation of the bounding box and presenting the score of the object with respect to Intersection of Union. Next, YOLOv5 (You Only Look Once) and U-Net models has been developed as efficient and accurate hybrid system towards detection and segmentation of the sickle cell anemia. In this U-Net Model is used to the segment the object detected by the Yolov5 technique. Further deep viii newlinelearning model is used to count the sickle cells