Stall Catchers: Play for Progress in Alzheimer’s Disease Research

As red blood cells squeeze through blood vessels, they can get clogged along the way; these blockages can be particularly detrimental in places like the brain. While this happens to some degree in healthy individuals, researchers observed that these blockages can significantly decrease blood flow in humans and animal models affected by Alzheimer’s disease. However, the underlying mechanisms and their connection to impaired cognitive function remain unclear.

This led Chris Schaffer, a physicist at Cornell University, to study blood flow at the capillary level in Alzheimer’s disease mouse models. Using a fluorescent dye, Schaffer’s team discovered that neutrophils caused red blood cells to stall in capillaries.¹ This blockage was treated with antibodies and resulted in improved blood flow, memory, and reduced cognitive symptoms in mice. However, analyzing millions of capillary images was time-consuming and existing algorithms were unable to identify stalls effectively. Schaffer said, “I just saw such a great educational and outreach opportunity by promoting this as a citizen science project. It was an exciting new finding that's tied to a disease people really cared about.”

In 2014, a mutual colleague introduced Schaffer to Pietro Michelucci, a cognitive scientist and the director of the Human Computation Institute, who sought an opportunity to combine the complementary strengths of humans and machines. Together, they developed Stall Catchers, an online game where volunteers peer through a virtual microscope and score blood vessels in mouse brain videos as “flowing” or “stalled.” These videos come from Alzheimer’s disease mouse models that were exposed to various conditions, including high-fat diets and exercise.^2,3 Multiple citizen scientists review each vessel to generate an average score.

Since its 2016 launch, over 130,000 participants, ages six to 88, have contributed. Stall Catchers also hosts “Catch-a-Thons,” global 24-hour marathons where teams come together. The collective effort helped speedrun the creation of a huge database of annotated videos of known flowing and stalled vessels.

More recently, this growing collection of training data has led Michelucci to introduce machine learning challenges to develop bots for parallel analysis. “The human answer was better than the bot answer,” explained Michelucci. However, the combined results were even better. This blend of perspectives gave researchers a stronger “wisdom of the crowd” effect and became a tool in their toolbox for higher quality and faster analysis.

For Schaffer and Michelucci, a key goal is to always provide a path for the citizen science community to stay engaged. They also make it a point to keep volunteers updated on the progress of their contributions by sharing their efforts in research papers and other developments.

Want to submit your own citizen science project?

Tell us about it