Automatic Semicircular Canal Segmentation of CT Volumes Using Improved 3D U-Net with Attention Mechanism

Figure 8

CT slices of our dataset. (a₁), (a₂), and (a₃) represent the 37th, 39th, and 41st slices of a sample in our dataset. (b₁), (b₂), and (b₃) are the corresponding labels; the areas where the semicircular canals are located are marked in blue. (a₂’) and (b₂’) are enlarged pictures of (a₂) and (b₂).