Crowdsourcing to gather connectomic data

[mrjiaruiwang]

So I guess models are severely restricted by poor data, and it can be quite time consuming to process data. Given that we are now collecting a lot more data with the prevalence of MRI, maybe there could be a way to speed up the process of data collection. Computational methods tend to not work well on noisy MRI images, but humans are great. So could there be a way for the average citizen of the world to contribute to this problem via a crowdsourcing platform? They're not trained scientists, but they only need to get 5% of the problem right because we can combine the response from thousands of participants. Is this already being done? How could it be done?

[ajulian3]

Do you mean generation of a connectomics data set that would be available on a public platform? Im not familiar with MRI techniques. Is there a standardized procedure or is there different methods across the medical community? The benefits for science would be incredible if this was possible. However, HIPA agreements and doctor confidentiality agreements make this task incredibly challenging. Consequently, information could possibly be acquired from countries who don't possess the same barriers to entry.

[mblohr]

That is a great idea, but yes, HIPAA might be a problem. Could the data be made generic enough/untraceable to patients?

[wrgr]

We can definitely anonymize patient data...that's not a problem. I'm not sure what you would do with MR data, though - try to find and manually trace fibers from tensors? Would people do better than humans? For EM data, there are projects like Parity: openconnecto.me/stem, and eyewire.org.

[adjordan]

What exactly would the average person be contributing?