Yeah, that's exactly the type of 'noise' I was referring to. Improper histological staining, poor imaging skills, etc. are all things that the human mind can immediately discover. The human mind is smart enough to not rely on the assumption that all random variables implicated are conditionally independent of each other. Computers, on the other hand, use algorithms rooted in these assumptions.