I don't think it can be done. The best I can come up with, and I know it is not necessarily a reliable test, would be to turn off all the breakers feeding that box (good luck figuring out that you have gotten them all), disconnect all wires within the box (label them all first, so that you can reinstall everything correctly), and measuring resistance from each wire to the box itself. If the box is metal, and if it is properly bonded, then there should be close to zero resistance between the box and the N-G bond point back at the main panel. So the resistance between a neutral conductor and the box should also be close to zero. When I say "close to zero," I mean in comparison to the reading you should get between a phase ("hot") conductor and the same box. With the circuit breaker open, the resistance reading should include the resistance of whatever loads are connected to that circuit. You will be reading the resistance of the wire from the point of connection within the box to the load(s), and the resistance of the load(s), and the resistance of the neutral wire(s) from the load(s) back to the main panel's N-G bond point, and the resistance of the equipment ground wire from that same N-G bond point to the box (which, as I assumed earlier, is bonded). That value should be noticeably higher than the resistance value you would get from a neutral wire to the box.
The only really reliable way to tell is to disconnect all wires within the box and within the panel (once again, label them all first, so that you can reinstall everything correctly), and test continuity from the panel to the wires inside the box. This, of course, requires you to use a test lead that reaches all the way from the panel to the box.
Welcome to the forum.