Deep Neural Network With Scaled Conjugate Gradient

I am replicating an article experiment. And during the training process, the trainscg algorithm was used to update the weights of a shallow network. However, now I need to add a convolutional layer and so I had to move to a deep neural network, using the "trainNetwork" training function. But, the closest update algorithm was stochastic gradient descent with momentum (SGDM). First, I needed to update my network weights using the scaled conjugate gradient. To later evolve the studies with other algorithms, since he, until then, presented the best performance. The question is, how to use the Scaled conjugate gradient to train a deep neural network.