Convolutional Neural Networks for Recognition of Lymphoblast Cell Images

Figure 1

Architecture of ConVNet. The CNN consists of seven layers. Layers 1, 2, and 3 implement feature extraction of cell images. Layer 4 transforms 64 extracted features into one-dimensional array of size 65536. Layer 5 maps 65536 inputs into 64 outputs. Layer 6 drops 50 percent of the 64 inputs at random. Layer 7 performs classification of 3 types of ALL subtypes.