I’m surprised that an MRI scan can allow a synapse-level reconstruction. So surprised that I suspect I’ve misunderstood (I can’t read the original paper — I probably wouldn’t understand it even if it wasn’t paywalled).
They did diffusion tensor imaging. What this does, using MRI, is determine the local anisotropy of water flow in each voxel. You assume that this anisotropy aligns with with the axis of axons, since they limit the diffusion of water across their axis (water diffuses along them). You can then use the principal directions of the diffusion tensor to estimate in what direction the water, i.e. the axons are "flowing" giving you an approximate picture of how axons connect different parts of the brain.
MRI is ~5 orders of magnitude less precise than EM. Not even close to cellular resolution, let along single axon resolution. You can only see axon tracts, where thousands of axons may make up one pixel.
MRI can’t do imaging at synapse level, even high powered on small animals. This has nothing to do with connectivity between neurons. It is more like they do correlations on correlations, and find mammals share some basic connection pattern (by correlation measurements). It might surprise those who don’t believe in evolution
It's a bit of a tough read for outsiders (which I am), but with some basic understanding of the brain and a few lookups I think the paper is actually pretty readable in the end, so I'm attempting a quick explanation.
There's a lot of interesting stuff to read about Connectomes[0] on wikipedia if you're curious about the mapping thing. I don't think they have an actual network map there, but in the initial pages of the paper they describe using MRI measurements to assess 200 areas they split the brain into ("normalized to 200 voxels per brain"), and they test their MRI methods by re-creating two tangentially related measurements which attach importance to geometric constraints of the brain.
They mention: "Note that, as tractography does not actually measure axons but rather axonal bundles and fascicles, our length estimate reflects the wiring length of the macro-scale network."
So it's more akin to getting a mapping of Earths (as in, each animal is one) and splitting it into 200 voxels, then looking at the density of fiber optic cables laid out, for many types of "Earths". So they don't see each strand but the bundles, which could mean that they don't know how much "data" actually goes through but they do know what bundles go where, which is more the point of connectivity.
Then they go on to notice that there's overall little change of density across the measurements (40% maximum over 4 orders of magnitude of volumes) even though there are different structures (some might have only one one dominating continent while others have two, for example), and that within each "family" of measurements there's consistency in measured density across their samples.
They also mention: "Results suggest that intra-hemispheric connectivity compensates for poorer inter-hemispheric connectivity, maintaining the overall connectivity."
Which I think is something that at least from a functional standpoint has started to be looked at for humans (in studies like this[1]) but I don't know if any other study has done such an analysis as systematically as these guys, and I don't think ANY other study has done so across so many species and orders of magnitude of brain volumes.
(I am happy to be corrected by any actually-knowledgeable person passing by, if they are so kind as to strike down any mistake)
Is MRI really that high-resolution now?