Skip to main content
. Author manuscript; available in PMC: 2019 Mar 20.
Published in final edited form as: KDD. 2018 Aug;2018:1226–1234. doi: 10.1145/3219819.3219974

Figure 4:

Figure 4:

Network architecture implemented on the ADNI dataset. The encoder has two max pooling layers with sizes of 2×2×2. We replace the regular deconvolutional layers in U-Net with variations of the proposed voxel deconvolutional layers in the decoder path.