Ending where we might have started... detecting the tissue.
(Identifying regions of interest to analyze can be tricky in itself, and another potential source of error. I tend to think it's best to at least start with manual annotating regions of interest...)