Long time ago Wheeler proposed an idea that there is only one electron, some kind of a closed(?) four dimensional knot in spacetime, and in one particular time we slice this knot with a plane and observe it as many electrons at different places in space. In this picture a positron is just this same electron, but moving in an opposite direction. The direction itself is arbitrary, by opposite I mean if we trace the whole trajectory and at the place where it intersects this plane again we already turned back.
So for three-dimensional spacetime we would always have an equal amount of electrons and positrons at one time, it shouldn't hold for four dimensions, as we actually observe, right?
The annihilation process would be just particle turning around in time. And we lose two time plane intersections at this moment for this particle knot. How would the photon's "knot" look like? Because there is no antiphoton, I imagine many trajectories for a photon, so there are actually many photons, they are just straight lines in spacetime. It should hold for all particles that are antiparticles itselves.
There is always some confusion of what C, P and T transformations actually do. In this interpretation, C is reflection of particle trajectories in time dimension , P - in space dimensions and T - corresponds to the choice of direction we follow along the trajectory. So CPT just reflects the knot and we traverse it in the opposite direction. It immidiately makes sense!
A very wild thought would be to imagine how these knots would move, because if they were fixed, it would be boring. But it can't change its topology, so everything would allow to change, even the past, but the overall structure would be the same, and it would be reflected on a small scale taking form of quantum effects, but the complex structures as we are wouldn't notice.