Dimorphic Anemia Identification using Image Processing

Microcytic macrocytic anemia also known as dimorphic anemia has often presented problems in the diagnosis of hematological disorders. This work uses sophisticated image processing and artificial neural network to make an inference about hematological parameters specifically red blood cell indices (MCV, MCH, MCHC), RDW, serum vitamin B12 and folate levels. Peripheral blood smears and Prussian blue staining was used to evaluate iron status and convolutional neural networks (CNNs) for determining the optimal features that exist in blood images for improved diagnostic accuracy. The results presented also highlight the importance of nutritional deficiencies, chronic inflammation and malabsorption in the pathogenesis of dimorphic anemia. This type of analysis adds up to solutions catering for diagnostic needs since is less costly making use of manual evaluations and this is important for people in situations where they have less access to resources.