Brain scans reveal simple 3D grid structure | Hacker News Reader

Brain scans reveal simple 3D grid structure

nih.gov

105 pointssignifiers16 comments

Very intersting find. Though every time I see 'simple 3D grid' in combination with the brain, I think 'scanner or reconstruction artifact'. For my PhD research I've worked with DTI (Diffusion Tensor Imaging) data which was very hard to sensibly interpret as 'connectivity': The visualized structures depended very much on the parameters and reconstruction method used. And it was hard to match the visualized structures to actual physical structures.

I wonder if they verified in some other way (ie, microscope) that the simple 3D structures were really there. If so, very exciting news!

They addressed these concerns in several ways - firstly, yes there is microscopy on corresponding sections demonstrating that fiber-crossings were resolved correctly. Second, DSI imaging was performed on mouse hearts, as myocardium exhibits a strong directional organization but without the crossings shown in brain (negative control).

Unfortunately the full article is paywalled, but the Methods supplement is freely available and discusses these issues.

http://www.sciencemag.org/content/suppl/2012/03/28/335.6076....

MRI is mathematical voodoo even when it's working - it's easy to fool yourself at any of several stages in the process. "New scanner gives 10x improvement in detail, shows amazing and unexpectedly simple result" is just screaming for verification.

But it needs it obviously enough and the claim is interesting enough that further verification (or falsification) should happen soonish, as research timescales go.

In fact, this group has already done very detailed validation in cat brains: in-vivo MR imaging followed by ex-vivo staining+microscopy for validation :

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2789885/

It is good to hear this, I had my doubts but no experience anywhere near MRI reading. Looking at those colourful tubes, it is not in the least bit obvious how much of it is pure modelling and how much is cleaned-up scanner output.

"New scanner gives 10x improvement in detail, shows amazing and unexpectedly simple result" is a pretty punchy-sounding headline, too. They should go with that next time!

Suggest changing the title to: "Imager Reveals Grid-Like Brain Connections Simpler Than Previously Believed."

Or, adapting the subtitle: Brain scans reveal simple 3D grid structure

Agree that the current editorialized HN title (Beautiful math: Human brain Connectome project) is nonsensical. The word "math" doesn't appear in the article at all, for example.

The particular finding in the article is just the most recent insight from next generation DSI computational modeling breakthroughs. The article highlights the hardware/device engineering, but the real news is that by using these new modeling techniques, nearly all modern MRI scanners can benefit, largely unmodified.

As someone who works at the intersection of informatics & device engineering, I find these visualizations stunning.

And the embedded video in the article is not just eye candy, it actually represents a class of diagnostic tools that didn't exist a couple of years ago.

Guilty as charged on a modest bit of editorializing; just google DSI or its related cousin Diffusion Tensor Imaging and tell me that's not pretty cool.

As for "no math" mentioned in the article, that is true. But a cursory search for the terms tensor or fractional anisotropy and weighted trajectory projection models will quickly reveal just how much math is involved, regardless of a press release that glosses over that fact.

For a good introduction to the Connectome project, this is excellent: http://www.thersa.org/events/video/vision-videos/connectome

And the "game" that came from it: https://eyewire.org

Would be cool if someone from like Zynga or other gaming company could actually help expand this to the general public.

They could start helping themselves by making it easier/possible people to figure out what it is all about without signing up or in.

The limited connectivity is very interesting -- my experience with artificial neural nets is pretty limited, but are there many well-studied ANN models that have a similar structure to this? The spatial restrictions are so simple that it seems like they'd lend themselves to efficient implementation.

Original article:

http://www.sciencemag.org/content/335/6076/1628.abstract [paywall]

Someone should start a project to document all of the mechanisms in cells somewhat resembling machine phase nanotech and call it the Cellular Doohickeyome Project. (Or is it just me, but aren't there an awful lot of -omes now?)

Does anyone know the resolution of this scan? Are there many complicated structures that are smaller than this apparent grid, or are we getting close to seeing "live" images of a fully wired, working brain?

I think that we are really starting to be able to model a lot of the details of the human brain and that is very exciting.

https://vimeo.com/23225093

Reply on news.ycombinator.com