I posed the following questions a couple of months ago and never received any answers. Based on what we have seen recently and a recent article from a few days ago, I think I can partially answer some of my own questions. Would love to hear any additional insights from the Stall Catchers and/or Research Teams.
Questions from July:
Question 1.) Biomedical Research question on Amyloid Plaques - I was under the impression that we had completed enough catching during the #CrushALZ dataset to enable the research team to answer the pending research question of "whether stalls occur closer to amyloid plaques than flowing vessels in the brain in Alzheimer's?" I thought Chris thought we would be able to answer this specific research question within a few weeks of completing the #CrushALZ competition. Lindsay, anything you can share on this? ( @lvinarcsik )
Answer: An article was published on Sept 8, 2017.
The conclusion from our initial dataset looking for a link between stalls and beta-amyloid plaques was:
"A first finding has shown no association between stalls and beta-amyloid plaques, toxic protein deposits in the brain, despite earlier suspicions of a link."
Question 2.) Pipeline and improved outlining for future datasets. Getting us the next set of data to begin answering specific questions regarding treatment-centered research was stated as being a priority. Any updates regarding the ability to get this improved data in a timely fashion? Will we have enough data to keep our stall catchers busy? Lindsay or Pietro any thoughts that can be shared? ( @lvinarcsik , @pietro )
Answer: New datasets are being provided. I think most of us would concur that the quality of the outlining algorithm has been significantly improved in the more recent datasets. Kudos to the IT/algorithm folks for making these much more accurate. Still get a few with no apparent vessel targeted, but much, much better.
Question 3.) Dynamic Consensus Methods. Pietro, We continue to crunch away on the CrushALZ dataset. Any updates on your team's progress in confirming your Dynamic Consensus statistical methods? Not trying to scoop any future publishable articles, but as my children used to say, "Are we there yet?" I would love to hear that we really may not need 20 community votes for every image as we move forward. Of course, we will then need more data quicker from the lab. Pietro, any additional updates you can share? ( @pietro )
Answer: It does appear that the new "stopping algorithm" (Dynamic Consensus) has been implemented. We have seen instances where we have processed all of the existing data in the dataset. The research team does appear to be introducing new data on a more frequent basis to keep our "Point Redemptions" alive and well. Thanks for getting new data to the catchers.
Would love to hear more on any of these topics (or others).