1st paragraph suggests using Neural Networks (you know, the thing that guesses the output?) to add detail to medical scans and security footage.
Ctrl+W
btw, the reason I clicked in, was because I was curious as to how an FPGA (i.e. stuff that ends up being more expensive for any one task, as compared to ASICs^[I'll consider a GPU as loosely an ASIC] in turn for being reprogrammable) managed to be cost effective.