You cannot resume a human mind from static imagery of a brain, any more than you can resume the apps running on your smartphone from static imagery of your phone's circuitry.
The FBI confronted this reality when trying to crack the San Bernardino shooters' iPhone.
The 2008 Whole Brain Emulation Roadmap seems to completely miss this point, except perhaps in its handwaving appendix on "non-destructive and gradual replacement". Those fantasies will eventually be realized, and only then will minds be able to be hibernated (and thus cheaply and quickly copied.)
So the Age of Em is extremely unlikely to happen in the manner and timeframe that the brilliant Robin Hanson expects.
There is in principle a way around this hibernation problem. You just have to emulate the entire development of a brain, and then feed it a suitable lifetime of input to train it into a desired state. This approach is computationally more expensive, and would require lots of slow (and morally objectionable!) iterations. Or you could try to bypass the iterations by instrumenting various (by definition unwilling) human subjects and log a few decades of their sensory inputs. Thus you'd only be able to emulate the, um, victims of your experiments, rather than emulating arbitrary cognitive superstars. Still, you'd be able to cheaply and quickly make copies of them, and an Age Of Emulated Boys From Brazil would then be possible.