Joint Dense Residual and Recurrent Attention Network for DCE-MRI Breast Tumor Segmentation

Figure 6

Dense residual module based on dilated convolution. The “Conv1 × 1” represents a convolution block composed of a 1 × 1 kernel size convolution, GN, and PReLU. The “Conv3 × 3” represents a similar block but utilizes a 3 × 3 kernel dilated convolution. The “d” represents the dilation rate. It should be noted that the “Conv1 × 1’ after “Conv3 × 3” does not contain a PReLU. To simplify the diagram, this has also been named “Conv1 × 1”.