Classification of White Blood Cells Empowered with Auto Encoder and CNN

Abstract

Differential counting of white blood cells (WBCs) is a well-established clinical practice for assessing a patient's immune system state. Information about our health state may be gained by determining the amount and type of white blood cells (WBCs). The quantity of white blood cells (WBCs) may be used to identify disorders such as leukemia, AIDS, autoimmune diseases, immunological deficiencies, and blood diseases. Convolutional Neural Networks (CNNs), in particular, have a tremendous impact on the medical industry, where a large quantity of pictures must be processed and studied. Images and objects may be categorized and identified using ACNN in this research. Input is provided in the form of raw pixels, and the algorithm outputs an indication of how likely it is that the pixels fall into one of many different categories. Convolution and pooling are added to each layer to minimize the parameter magnitude by a significant amount. The proposed approach will take images automatically from data set and reduce the size of images with auto approach for working faster. It is time-consuming and exhausting to manually locate, identify, and count the many WBC subclasses. Accuracy in classification and counting is directly related to the skill and knowledge of the workers