Apply BatchNorm, Relu 3x3Conv2D, optional dropout :parameter x: Input keras network :parameter concat_axis: int -- index of contatenate axis

Build a denseblock where the output of each conv_factory is fed to subsequent ones :parameter x: keras model :parameter concat_axis: int

load MURA dataset

Run MURA experiments :parameter batch_size: int -- batch size :parameter nb_epoch: int -- number of training epochs :parameter depth

Apply BatchNorm, Relu 1x1Conv2D, optional dropout and Maxpooling2D :parameter x: keras model :parameter concat_axis: int -- index of contaten

Build the DenseNet model :parameter nb_classes: int -- number of classes :parameter img_dim: tuple -- (channels, rows, columns) :pa

Build a denseblock where the output of each conv_factory is fed to subsequent ones. (Alternative of denseblock) :parameter x: keras model :pa