Semantic segmentation using deeplabv3+ with resnet 18

Hello, good time

I have a question related to the topic of semantic segmentation with Deeplabv3+ Network with the weights of a CNN network already trained in MATLAB, can we remove a layer from the network structure or add a dropout to it?

Because it does not allow me to delete it at all, but I was able to replace Relu layer with Leaky Relu. Now I wanted to know if it is possible to remove layers from the structure?