FIGURE 1.

The structure of the three-dimensional (3D) multi-task deep learning model. The proposed network included three modules, (1) shared feature extraction module, (2) glaucomatous optic neuropathy (GON) classification module, and (3) myopic features (MF) detection module, respectively. It was built based on ResNet-37 network with 3D convolutional layers and global average pooling layer. The input was an OCT volumetric scan of size 200 × 1000 × 200 pixels after image pre-processing and the output was Yes/No GON and Yes/No MF.