First and foremost, this presumes that the brewery uses a pure/single-strain culture, and that the bottling strain is the same strain as the fermentation strain. If any of those presumptions are incorrect, then you probably aren't going to get what you're looking for.
I haven't built starters from petri dishes too often, but I would look for an average-sized, light cream colored, waxy/shiny colony to load my loop from.
Personally, when I'm trying to build up a culture from bottle dregs, I make 100mL of starter wort from DME and add the dregs, plus about 100mL of the beer to the starter. This ends up with a half-strength starter mixed with a half-strength beer. The lower starter gravity is a bit easier on the yeast. The beer helps reduce the pH of the starter, and contains some alcohol and hop compounds. My thought is that this will mitigate the growth of any potential spoilage organisms while waiting for the yeast to wake up.
The beer will also bring over any yeast still in suspension, rather than only those that have flocced out. In the end this is kind of the opposite of selecting for a pure culture, since you're selecting both the most flocculant and least flocculant variants from a culture instead of one specific colony, but I've always felt like this has a better chance of replicating the performance of the original culture than selecting just from the dregs.