An idea?
I you wrapped the suspect device in say 1/4 or 1/2 inch fiberglass (or similar material structure)you would get a different infared temperature where the leak was vs. where it wasn't.
Why? Where no leakage existed there would be only conduction heat at the outside surface of the "fiberglass" and where a leak was present there would be leakage mass heat transfer in addition to conduction heat transfer.
Different surface temperatures = leak detected.
If patentable, I claim the rights.