A New Abnormal Red Blood Cell Data Set

Blood diseases a worldwide public health problem such as thalassemia, anemia malaria. It diagnosis basis in abnormal shape,size and color of red blood cells (RBC). in this paper the data were obtained from 1000 RBCs that extract from 150 blood smear images for different diseases. After that, several image pre-processing steps done to extract statistical features using a new system. k-means well-clustering algorithm was used to solve the challenges of the color variability of both the background and the cell's pixels is the main problems we met in this work. 100 features for 30 types of normal and abnormal RBCs. The anew data set was test with different machine learning classification algorithms. The result show very high accuracy more than 93%, and evaluation algorithms models achieve more than 94.4% These result conclude that the algorithms have high ability of distinguish between the true negative and true positive more than 95% and Root mean squared error lease than 0.38, the fact that the features are very accurate in describing RBCs.