Figure 7.
The structure of the MVC-net [67]. MVC-net consists of three parts: a convolutional layer for feature extraction (the convolutional layers are connected with X modules for feature joint learning); a spine landmark output layer and a Cobb angle output layer.