In the liver and tumor segmentation model, it's possible to approximate the Brain Segmentation workflow with RepeatChannelD to enable usage of ConvertToMultiChannelBasedOnBratsClassesd. However this results in greater, unnecessary memory usage.
Adding a DeleteChannelD transform would allow the duplicated channels to be pared down before feeding into the network.