CMMCSegNet: Cross-Modality Multicascade Indirect LGE Segmentation on Multimodal Cardiac MR

Figure 5

Comparisons of the loss weights of different cascade prediction in proposed CMMCSegNet, where the LSGAN adversarial loss and vgg perceptual loss are employed and the loss weight parameters are manually given. (a) the LGE image and the corresponding ground truth; (b) from left to right: the predicted results under different loss weights , , , and , from top to bottom: .

(a) LGE, fake bssFP, and ground truth

(b) CMMCSegNet predictions in different loss weights