Comparative analysis of White Blood Cells Classification using Deep Learning Architectures

TheWhite Blood Cells (WBCs) are essential for the body's protective components of our immune system. To monitor a person's health status, WBCs are used. The variation in WBCs count in blood cells leads to hematological disorders. The exact diagnosis of variation in number of WBCs can prevent different types of disease. The illness includes immune system infections, including anaemia and leukaemia. A doctor must identify the number of WBCs in man's blood and classify the WBCs to avoid the disease by a more experienced clinician. The architectures of the Convolutional Neural Network (CNN) in deep learning are AlexNet, LeNet, and Visual Geometry Group Network (VGG-16) are embedded with softmax classifier. The architecture of CNN in existing work is modelled using python which provides minimum accuracy. Features are extracted from images using the proposed CNN framework with the help of two convolution layers. Features are fed into a completely linked blood cell classification layer using CNN, VGGNet and GoogleLeNet, to classify the types such as lymphocytes, monocytes, eosinophils, basophils and Neutrophils. Experiments on standard benchmark data sets are performed to estimate the classification accuracy of blood cells by the suggested system. As a result it was observed that the GoogLeNet with the accuracy of 93.43 and 91.72 using Relu and LRelu activation function overlooks comparisons between CNN, and VGGNet