I've done something similar in the lab but had all the VM's powered down before taking that point in time VM backup. If you try to do this on a running ISE VM, the required snapshot will hang the ISE application forcing you to reboot it. I would not recommend it for a production set up.
Doing a parallel upgrade like you are wanting is common though. Take a backup from the production 2.2, restore it to a 2.2 or 2.4 VM, and it becomes a running working admin node for your new upgraded deployment. You can deploy new VM's for your deployment in parallel which means the old deployment is always there to roll back to if something goes wrong. Multiple ways to go about this, but preserving the old VM's is just a matter of turning them off, and deploying a new 2.4 vm in it's place with the same IP. It's some extra work but certainly doable, added benefit of doing everything in a supported manner which the cloning idea would not be. Assuming ISE was designed and deployed in a HA fashion, there should be no noticeable impact to endpoints.
