Cerebral Arterial Stenosis Detection Based on a Retained Two-Stage Detection Algorithm

Figure 3

Structure of deformable convolutional network. The deformable convolutional layers were represented in this study by adding 18 output channels to a 2D convolution layer with a convolution kernel size of 3 × 3, a convolution step of 2 × 2, and a feature map padding of 1 as offsets before a 2D convolution layer with a convolution kernel size of 3 × 3, a convolution step of 2 × 2, and a feature map padding of 1.