Figure 3.

Schematic representation of the proposed (CM-YNet) architecture showing how the total loss function is computed. The segmentation and parametric convolutional neural networks (CNNs) are integrated in the Y-Net model described above. At inference, the output of the segmentation network ${\widehat{\mathbf{y}}}_{\theta }\left(\mathbf{x}\right)$ is used to obtain a new prediction.