Hacker News new | past | comments | ask | show | jobs | submit login
Mapping the Brain at High Resolution (mit.edu)
78 points by ArtWomb 34 days ago | hide | past | web | favorite | 11 comments



Their resolution is 60 x 60 x 90 nm which is far below the 20 um that BigBrain used in 2013 for their whole-brain scan [1].

Especially, axons are 0.2-20 um in diameter. With a resolution of 20 um you won't really find them, but with 0.01 um, you will certainly be able to identify them. This could lead to new anatomical insights in human brains. I remember recently someone complaining on hn that a new finding about a connection of two brain regions in mice hasn't been confirmed in humans yet [2]. If scans of entire human brains at 0.01 um resolutions were available, you could verify claims like these by just checking the data.

Edit: corrected the BigBrain resolution.

[1]: https://en.wikipedia.org/wiki/BigBrain

[2]: https://news.ycombinator.com/item?id=18943385


You shouldn’t mix units like that.

Everything should be experessed in nanometers only, and avoid micrometers, since expressing nanometer sizes in micrometer units requires confusing decimal points, and an extra cognitive load to translate fractional proportions.

  Axons are 200 to 2000 nanometers in diameter.
Much easier to read and compare to the nanometer resolutions defined in the first paragraph.

Anyway, we’d need thousands of desireable human brains to draw conclusions. And my desireable, I mean smart, sociable people mostly. You’ll want total morons in the data set too, but only to distinguish total assholes in the training data, when we use AI to rationalize their extermination. That way, the Hitlers go to the concentration camp, this time.


The Wikipedia link for BigBrain you provided lists the resolution at 20 µm isotropic 3D voxels. That is ~2.71 µm in each direction.


Nope. Yes, the Wikipedia link says 20 um isotropic resolution, but that means 20 um per direction. 20 um is a unit of length, while 20 um ^ 3, a unit of volume. If it were about volume, you'd be correct.

Also, I just looked at the paper [1] and here it says:

> To generate a data set with isotropic resolution, we down-scaled all images to 20 µm by 20 µm to match the section thickness of 20 µm.

So that confirms the 20 um per direction, no?

Edit: oh no, I just see that I wrote 1 um above, not 20. Not that 2.71 is correct but I guess that's what confused you. Corrected. Thanks!

[1]: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.899...


<slaps forehead> You are right about the resolution. Apparently I did not have my coffee because I work with voxels all the time.


It's all good :)


I can't figure out what is meant by "expand" in this and other quotes:

"The new technology combines a method for expanding brain tissue, making it possible to image at higher resolution, with a rapid 3-D microscopy technique known as lattice light-sheet microscopy. In a paper appearing in Science Jan. 17, the researchers showed that they could use these techniques to image the entire fruit fly brain, as well as large sections of the mouse brain, much faster than has previously been possible."

--> I mean, are they pumping the tissue up like a balloon or what?



Considering all 302 nerouns and 7000 synapses in c. elegans has been known for a decade. But how this maps into c. elegans behavior is not understood.

Multiply this by a hundred million to get mammalian complexity


Not only that, the entire organism is only about 1000 cells. (Source: https://en.wikipedia.org/wiki/Caenorhabditis_elegans)





Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: