Hybrid KGA algorithm For Clustering RBC’s Morphological Abnormalities

This paper proposes a new hybrid algorithm (KGA) of two known algorithms for the data clustering process. These algorithms are the Genetic Algorithm and K-means. In this paper, a comparison of the hybrid KGA algorithm’s performance with the original algorithm’s Genetic algorithm and K-means. The comparison was based on clustering Red Blood Cells morphological abnormalities features. The RBC’s (red blood cells) morphological are represented through three types of features: Shape descriptor features, Shape signatures features and wavelet features. Through this paper, a study of the three features set was to find which was more informative of the RBC’s morphological abnormalities. The extracted features obtained from blood film images collected from two known hospitals in Gaza strip. The features with different sizes, processed and normalized. The results Demonstrate clearly that the shape descriptor features were the best, with GA 99.48%, K-means 83.16% and KGA 99.56%.