search

open-connectome-classes / StatConn-Spring-2015-Info

introductory material
18 stars 4 forks source link

Statistical Combination of Data #43

Open mrjiaruiwang opened 9 years ago

mrjiaruiwang commented 9 years ago

There are a lot of methodologies for collecting data from the brain, and I'm wondering if it would be useful or possible to combine the data from all these techniques to form one, unified version of the "brain map". That way we can be sure that the data we work on is consistent and reproducible and generalizable.

Side Note: As much as I like to use theory and computation for solving my problems, I personally like to stay close to how data is collected so we can be aware of what types of assumptions we should and shouldn't be making.

jovo commented 9 years ago

how would you combine EM data (nanometer resolution) with MR data (millimeter resolution)?

On Tuesday, February 3, 2015, mrjiaruiwang notifications@github.com wrote:

There are a lot of methodologies for collecting data from the brain, and I'm wondering if it would be useful or possible to combine the data from all these techniques to form one, unified version of the "brain map". That way we can be sure that the data we work on is consistent and reproducible and generalizable.

Side Note: As much as I like to use theory and computation for solving my problems, I personally like to stay close to how data is collected so we can be aware of what types of assumptions we should and shouldn't be making.

— Reply to this email directly or view it on GitHub https://github.com/Statistical-Connectomics-Sp15/intro/issues/43.

the glass is all full: half water, half air. openconnecto.me, we're hiring! https://docs.google.com/document/d/14SApYAzxF0Ddqg2ZCEwjmz3ht2TDhDmxyZI2ZP82_0U/edit?usp=sharing , jovo.me, my calendar https://www.google.com/calendar/embed?src=joshuav%40gmail.com&ctz=America/New_York

mrjiaruiwang commented 9 years ago

Imagine if EM data could be reduced down to a sum of gaussians, where each gaussian would be like a "pixel", the most elementary unit of measurement. Then reduce MR data in the same way. I think a good start would be to try to align these two datasets according to some prior, then find the maximum likelihood estimate "pixel" that is itself gaussian distributed with mean and variance some function of the two EM/ER means and variances.

I don't actually know how EM and MR data are collected or the underlying principles though, so I'll have to look into that first.