9 ms·

<shameless_plug> "Jülich Katrin Amunts has begun work on a detailed atlas of the brain which involved slicing one into 8,000 parts which were then digitalized

by nkassis 15y ago

<shameless_plug> "Jülich Katrin Amunts has begun work on a detailed atlas of the brain which involved slicing one into 8,000 parts which were then digitalized with a scanner." The lab I work for is working with them on this project. Basically those slices are taken at 10 microns and we are working on fixing some of the problems (when you cut a brain that thin you end up with a lot of tears and deformations) Then our lab is reassembling the slices (with algorithms) realigning the slices so you can cut through the slices in any directions and get the cross section of those slices and look at them. Realignment of the 8000 slices can take weeks or more (depending on how many cores you can run the stuff on) and is a rather complex process. Here a link for those wanting more info: https://cbrain.mcgill.ca/tools/visualization https://cbrain.mcgill.ca/tools/visualization Scroll to the bottom (well look at the whole page, brainbrowser is what I've been working on). You can see a image of the interface and some of the slices. I'll try to find a video of the project. </shameless_plug>

apl 15y ago
Impressive. > Basically those slices are taken at 10 microns and we are > working on fixing some of the problems (when you cut a > brain that thin you end up with a lot of tears and > deformations) Can you talk about how you deal with the inherent deformations? I've spent hundreds of hours on cryostats preparing rodent brain slices for autoradiography, and getting usable ones at 15-20 microns is a painful process. Should be even worse if you need to retain morphological information...
- nkassis 15y ago
  I'm probably not the best person to explain the whole process but I can put you in contact with the person who's working on it. (My email is in my profile). I'll try to explain what I know: For the most major tears and breaks we've actually had to fix them ourselves (Some were quite awefull, for example part of the right hemisphere on one slice broke off and was put back upside down ;p). For the less major thing, they can be fixed by looking at the surrounding slices, this is also how the slice are realigned, it's an iterative process where you take chunks of slices and aligned them then increase the chunks etc. At least that's what I gather. (Software developer talking about Brain imaging stuff is a bad idea ;p) EDIT: Also as for the morphological information I'm pretty sure that some of it will be affected by the process but I don't know how much. Our interface will provide the original and modified slices and a heat map of changes for comparaison in those cases. I don't know how that can be avoided.
DocSavage 15y ago
There are several orders of magnitude fewer neurons in the fly vs humans. For the fly, we are reconstructing neural processes using EM slices that range from 50 nm to 7 nm in thickness. At that scale, we can reconstruct neural processes and actually see synapses -- just because neurons are in proximity doesn't mean they are actually connected in the biological sense. Since I'm not as familiar with human neurons, could you tell me at 10 microns, can you resolve the axons, dendrites, and more importantly the synapses? Any references to papers would be great :)
9ec4c12949a4f3 15y ago
Once you have a great map of the structure and chemical process at one given state, how do you intend to map out the processes that take place? Or is that way beyond the scope of your project?

by nkassis 15y ago

apl 15y ago

Impressive. > Basically those slices are taken at 10 microns and we are > working on fixing some of the problems (when you cut a > brain that thin you end up with a lot of tears and > deformations) Can you talk about how you deal with the inherent deformations? I've spent hundreds of hours on cryostats preparing rodent brain slices for autoradiography, and getting usable ones at 15-20 microns is a painful process. Should be even worse if you need to retain morphological information...

nkassis 15y ago
I'm probably not the best person to explain the whole process but I can put you in contact with the person who's working on it. (My email is in my profile). I'll try to explain what I know: For the most major tears and breaks we've actually had to fix them ourselves (Some were quite awefull, for example part of the right hemisphere on one slice broke off and was put back upside down ;p). For the less major thing, they can be fixed by looking at the surrounding slices, this is also how the slice are realigned, it's an iterative process where you take chunks of slices and aligned them then increase the chunks etc. At least that's what I gather. (Software developer talking about Brain imaging stuff is a bad idea ;p) EDIT: Also as for the morphological information I'm pretty sure that some of it will be affected by the process but I don't know how much. Our interface will provide the original and modified slices and a heat map of changes for comparaison in those cases. I don't know how that can be avoided.

DocSavage 15y ago

There are several orders of magnitude fewer neurons in the fly vs humans. For the fly, we are reconstructing neural processes using EM slices that range from 50 nm to 7 nm in thickness. At that scale, we can reconstruct neural processes and actually see synapses -- just because neurons are in proximity doesn't mean they are actually connected in the biological sense. Since I'm not as familiar with human neurons, could you tell me at 10 microns, can you resolve the axons, dendrites, and more importantly the synapses? Any references to papers would be great :)

9ec4c12949a4f3 15y ago

Once you have a great map of the structure and chemical process at one given state, how do you intend to map out the processes that take place? Or is that way beyond the scope of your project?