Classification of White Blood Cells Subtype Using MCNN

Abstract

White Blood cells are building blocks of the immune system of humans as they fight different types of infections, which is vital for healthy recovery. Changes in the number of White Blood Cell subtypes (WBCs) rule out certain diseases such as infection, heart disease and diabetes in medical practices. Conventional methods of counting the number of WBCs are dependent on manual testing and have chances of human error and the automated method apparatus is very costly. Thus the classification of White Blood Cell subtypes is of vital importance. In this study, CV based solution is proposed for White Blood Cell subtype identification. Different MCNN-based models along with transfer learning-based models (VGG16 & Resnet50) are trained and implemented for performance comparison and the effect of different training parameters on the performance of the models is also explored in the study. It was observed that changing the training parameters also affects the accuracy of the model. The highest accuracy of 96.6% was achieved using the MCNN-based model for the classification of White Blood Cells.