Clicked on this expecting to read about an AI enhancement technique. Was very refreshing to read about actual hardware improvements.
They just used the MRI to map a high-resolution image obtained via light sheet microscopy. I doubt you can do that with living animals. Without slicing the brain that is.
The title here is highly misleading.
> The diffusion tensor images (DTI) @ 15 μm spatial resolution are 1,000 times the resolution of most preclinical rodent DTI/MRI. Superresolution track density images are 27,000 times that of typical preclinical DTI/MRI.
The resolution of the raw MRI images is significantly higher, without that increased resolution it would be impossible to align the light sheet images. The light sheet images are not use to "improve" MRI resolution.
The highest resolution MR images I have (yet) obtained were ~20 µm^3 voxels on ex vivo (human) tissue samples fixed in agar with Gd3+ as a dopant, scanned at 12 T on a preclinical scanner. The coupled vibration of the gradient set causing blurring in the image domain at the extremities of the FOV was the limiting factor. I recall I did a partial Fourier acquisition – as ultimately we were limited by both T2* and vibration – and ended up trying to do POCS on a 4096^3 dataset and just needing tons and tons of ram to do it over something useful, like a weekend. Happy memories.
Think of it like how you might be able to reverse engineer and recreate a microchip from high enough resolution imagery of an existing microchip.
One thing that the MRI studies don't address are the types of synaptic connections. Neurons aren't all just excitatory of inhibitory. There's a massive amount of modulation happen with numerous types of neurotransmitters and other signaling molecules.
We've had the complete C. elegans connectome for 30+ years, and know very little about how it actually generates behaviour... because synapses are only a small part of the picture.