Detection of Sickle Cell, Megaloblastic Anemia, Thalassemia and Malaria through Convolutional Neural Network

This paper presents an alternative method to diagnose Malaria and Anemia (Sickle Cell Anemia, Megaloblastic Anemia and Thalassemia) as well as to differentiate between them. First, different related high resolution images of blood samples are taken from multiple datasets. Second, Convolutional Neural Networks (CNN) technique is implemented and applied in order to process the images without the need of the standard protocol of Complete Blood Count (CBC) test. The implemented convolutional Neural Network has been designed using Python to train on a number of microscopic images. After completing the training phase, the built model has been tested on other images to classify them into normal blood cells, Malaria, Sickle cell anemia, Megaloblastic anemia or Thalassemia. Third, the diagnosis is made based on the outcomes. Finally, the accuracy of results is assessed. The total accuracy of the test is 93.4%. The suggested approach yields promising outcomes that help diagnose blood samples faster, with low cost as well as without the need of an analysis laboratory.