Blood Cell Segmentation and Classification by Machine Learning

Abstract

The immune system is the third defensive line of the human body which defends the body from viruses, bacteria and pathogens. This natural protection detects and kills defective cells like the tumor cells. The immune system contains immune organs, immune and immune cells. The essential constituent of immune cells is white blood cells (WBCs), and they are an essential part of our body immunity. The light disperses system theory is used by automatic machines to compute red blood cells (RBCs) and WBCs. WBCs can be classified into five distinct types: basophils, eosinophil, neutrophils, lymphocytes and monocytes. They are granular in the first three forms and non-granular in the second two. It was ineffective to differentiate these types through the use of the light dispersal process. In this paper we will provide data that will consist images and .xml file of each image then system will provide us the information of image. We will segment out the blood cells images based on .xml file information. Also, will extract the features of blood cells depending upon the nature of image. To extract the color features will use Grid Color Moment Algorithm, for texture features Local Binary Patterns Algorithm, for classification will use K Nearest Neighbors Algorithm. Finally, will calculate the accuracy from our given data.