Response to 'Detecting (Positive) Cells with QuPath'

[@LéoLeplat]

When unlocking and resizing an annotation containing detections (like at 11:15), wouldn't it be better if QuPath was automatically resolving the hierarchy around the annotation? Meaning that all detections located below the new area would automatically become children of the annotation, and the rest of the detections wouldn't.

[@petebankhead]

That happened in QuPath v0.1.2 and earlier. The trouble was that it could be a massive performance issue: imagine having a huge and complex tissue region containing holes and disconnected pieces, and >1 million cells. On every move for the outer annotation, QuPath would have to determine the relationship between all cells and that region... and potentially other annotations that could be inside it (which is even more expensive). Switching to make resolving the hierarchy a separate command meant that the software could be far more responsive in the majority of cases where resolving isn't needed. I wrote about this and related changes at petebankhead.github.io/qupath/2019/11/17/changing-the-hierarchy.html