Geometric Feature-Based Classification of Anemia in Blood Smear Images using Decision Tree

A Machine Learning Technique that includes a set of procedures to analyze blood smear images, like preprocessing, feature extraction, Red Blood Cell classification, and results. Analysis of cells in the blood smear image is important before classification. In this approach, It analyzes blood smear images taken from a microscope which has 1000x magnification. It involves preprocessing techniques like conversion to grayscale, Gaussian blur, conversion to binary scale and removal of noise, removing intersected cells, and small cells. This study finds the particular kind of red blood cells present in the smear, which may be normal rbc or any type of anemia. Considering that abnormality in the red blood cells indicates the associated blood anemia. The primary attention was on the process of classifying types of cell abnormalities using the Decision Tree classifier. This method counts the predicted labels for the corresponding features and gives output as maximum predicted label of abnormal cells (elliptocyte, schistocyte etc..). The presence of various anemia types and their respective percentages indicate the type of blood anemia.