Hey! So, we store the whole slide. In terms of deltas, we found the easiest way around it was a combination of tracking the slide through it's markup (OpenXML s

Hey! So, we store the whole slide. In terms of deltas, we found the easiest way around it was a combination of tracking the slide through it's markup (OpenXML sucks, but you can extrapolate some UUIDs which help with tracking.) A lot of our analysis actually comes from the images themselves though, as they are an easier format (I.e. to help us see if the slide has changed.)