WBCs-Net: type identification of white blood cells using convolutional neural network

On monitoring an individual's health condition, White Blood Cells play a significant role. The opinion on blood-related disease requires the detection and description of the blood of a patient. Blood cell defects are responsible for numerous health conditions. The conventional technique of manually visualizing White Blood Cells under the microscope is a time-consuming, tedious process and its interpretation requires professionals. There are significant medical applications for an automated method for detecting and classifying blood cells and their subtypes. This work presents an automatic classification method with the help of machine learning for blood cell classification from blood sample medical images. The proposed method can identify and classify the function of each segmented White Blood Cells cell image as granular and non-granular White Blood Cells cell type. It further classifies granular into Eosinophil, Neutrophil and non-granular into Lymphocyte, Monocyte in various forms. Because of its high precision, the proposed framework includes a neural network model to detect white blood cell types. To improve the accuracy of multiple cells overlapping and increase the robustness, data augmentation techniques have been used in the proposed system. Which has improved the accuracy in binary and multi-classification of blood cell subtypes.