How I understand them to be is that Elements = everything. Network Node elements are all the nodes you are monitoring, volume elements is a count on all the volumes (RAM, CPU, etc) you're monitoring, and Interface elements is how many interfaces you're monitoring. I think that it also takes into account NCM, UDT, and any other things like that that you have installed, but I'm a little foggier on how those are counted with the polling.
I know that to help with our polling that we don't monitor as many interface elements as are discovered. I only keep ethernet interfaces along with a few other choice ones that we need monitored in our environment. When it comes to volumes we monitor, I usually don't monitor those unless it is a server, router, or something else that we have deemed worthy of monitoring its disk usage. We have some scripts that prune out interfaces we don't want since we have so many.