Fig. 3From: Automatic quantification of superficial foveal avascular zone in optical coherence tomography angiography implemented with deep learningGraphical representation of the proposed deep learning network. The proposed deep learning network included an encoder and decoder. The encoder comprised two Conv-BN- ReLu blocks (C1–C2) and five pooling blocks (P1–P5). The decoder comprised five upsampling blocks (U1–U5) and a reconstruction block (R1). The output of each layer is a three-dimensional feature map of size (h × w × d), where h and w are the height and width of the feature map, respectively, and d is the feature dimensionBack to article page