Computer aided detection of abnormal red blood cells

Red blood cell classification and counting plays a very important role in detecting diseases like iron deficiency anemia, vitamin B12 deficiency anemia etc. In this research we intend to develop a standalone application that can classify the red blood cells into four abnormal types namely elliptocytes, echinocytes, tear drop cells and macrocytes. We will also provide the total red blood cell count. Thirteen Geometric features have been used to classify the red blood cells into the four abnormal types. We have used two data mining classifiers namely Artificial Neural Network and Decision Tree Classifier and we have compared the results of the two classifiers with respect to accuracy in classifying the red blood cells. The proposed method exhibits an accuracy of 95.27% for detecting elliptocytes, 96.06% for echinocytes, 85.82% for tear drop cells 85.82% for macrocytes and 89.76% for normal red blood cells.