Figure 5 | Scientific Reports

Figure 5

From: Deep learning network for integrated coil inhomogeneity correction and brain extraction of mixed MRI data

Figure 5

Topology of the pix2pix network. The topology and parameters of the (A) U-Net and (B) PatchGAN classifiers with two input terminals. The job of the PatchGAN classifier determines whether the two input images are the same pair and the input of the right terminal is the target (real) or prediction (fake). In the first three layers of Conv3DTranspose, dropout can prevent overfitting. “BN” is abbreviation of “BatchNormalization”. (C) Framework of the cGAN. The input “c” and “x” represent raw images that are affected by intensity bias and images with bias field correction, respectively.

Back to article page