Sickle cell identification using image processing and red blood cell morphological characteristics

Blood is the most vital liquid that helps sustain the healthiness and life of the human body. Biologically, normal red blood cells have circular shapes and play a key role in transporting oxygen and nutrient to tissues. Red blood cells are bendable, which allows them to pass through the veins and arteries smoothly. Sadly, there are exceptional individuals with abnormal blood cells call sickle cell disease. The physical shape of the abnormal blood cells is in sickle/crescent form. Sickle cell disease is hereditary, and a person becomes affected if at least one of the parents has the abnormal haemoglobin S gene. The danger of sickle cells is that they inflict many severe health conditions such as pain, tiredness, jaundice, kidney problem, and other critical illnesses. For many years, managing and diagnosing sickle patients is performed by collecting blood samples to manually observe the irregular shapes of the red blood cells using a microscope. This process is time-consuming and results in errors for large samples of blood. In this thesis, a compelling image processing method is proposed to optimize the detection of abnormality in human blood cells with the deep learning technique. Ten images of red blood cells were randomly collected from the online source using the Google search engine. Each image was analyzed using MATLAB codes for image processing using the blood cell area, eccentricity, diameter, extension, and form factor as input parameters. The study results show that the proposed technique has 71 – 100 percent accuracy, far higher than what is obtainable in the manual method. This technique can serve and enhance the current manual method of sickle cell segmentation because it is faster and more accurate