Figure 4
From: Spinal cord gray matter segmentation using deep dilated convolutions
Architecture pipeline overview. During the first stage, input axial slices are resampled to a common pixel size space, then intensity is normalized, followed by the network inference.
