Classification of White Blood Cells by Convolution Neural Network in Lens-Free Imaging System

Abstract

Over the past decade, the lens-free imaging technique has been considered a good way to reduce the volume and the cost of cell analysis tools. However, limited by lens-free optical amplification, the cell imaging not only has low resolution but also has diffraction phenomenon in lens-free system. Therefore, there is a major problem, which traditional methods can hardly classify diffracted cell images in the system. At present, the state-of-the-art algorithm in image classification is to use the convolution neural network (CNN). Fortunately, the training of CNN method is fully accordant with the application requirements of classification of white blood cells (WBCs). In this paper, we proposed a technique for WBCs classification based on CNN in the lens-free imaging system. According to the test, the accuracy of this method for WBCs classification can reach to 90%, and it has a very broad application prospect in point-of-care testing.