Detection of RBCs, WBCs, Platelets Count in Blood Sample by using Deep Learning

A whole blood cell count is an essential check in scientific diagnosis to assess common fitness conditions. Blood cells are traditionally counted manually using a cell counting chamber (hemocytometer) in conjunction with various laboratory compounds and solvent (chemical) compounds, which is time-consuming and very dull to experiment with. In this work, a machine learning system mastering the method of automated computerized counting of blood cells is proposed. By using machine learning and deep learning techniques, the blood cells and their counts can be identified with the best accuracy compared to the other existing techniques. The CNN is used for the image classification. One of the best techniques for achieving the best accuracy in the least amount of time for the blood cells dataset is VGG-16 technique. The proposed system is a combination of the CNN and VGG-16 methods. The learnt models are generalized for the training and testing of the different datasets. In general, this computer-aided system of detecting the blood cells is more useful for practical applications. According to the results of this research, the accuracy of counting blood cells is in the range of 90 to 95 percent.