Skip to main content
. 2022 Jun;12(6):3151–3169. doi: 10.21037/qims-21-846

Figure 6.

Figure 6

Synthetic T1WI and T2WI by DRUNet-101 with MSE loss, BCE loss and contextual loss. (A) Top: synthetic T1WI by MSE loss; bottom: synthetic T2WI by MSE loss. (B) Top: synthetic T1WI by BCE loss; bottom: synthetic T2WI by BCE loss. (C) Top: synthetic T1WI by contextual loss; bottom: synthetic T2WI by contextual loss. (D) Top: real T1WI; bottom: real T2WI. DRUNet-101, double ResNet-U-Net with double 101 convolutional layers; MSE, mean squared error; BCE, binary cross-entropy; T1WI, T1-weighted images; T2WI, T2-weighted images.