Scientists have created an unprecedented high-resolution map of the brain that reveals structures as small as those found in individual nerve cells.
They produced the 3D map from a compilation of images taken with nanoscale resolution, making it possible to pick out features measured in millionths of a millimetre.
Researchers plan to use the tool to study the abnormal connections between brain cells that underlie debilitating neurological disorders such as schizophrenia and depression.
“We’re talking about imaging close to the level of a molecule,” said Narayanan Kasthuri, a neurobiologist at Harvard University, who led the team behind the map.
Kasthuri and colleague Jeff Lichtman’s team built a system that automatically slices a subject brain into thousands of thin sections. After staining the slices to pick out different tissues, an electron microscope was trained to take a pictures of each slice. A computer then assigns different colours to individual structures and knits the images together to produce a 3D map.
The scientists demonstrated the power of the system by imaging the area of a mouse brain responsible for sensory perception. The results are published in the journal Cell.
Traditional brain imaging techniques, such as MRI, are straightforward to use, but can only resolve features down to about a millimetre. A German anatomical atlas called “BigBrain", resolves features of the human brain down to micrometres – thinner than a human hair – and almost on a scale of individual cells.
En contraste, Kasthuri’s tool produces nanoscale images of individual brain cells, their contents and their connections. “One pixel on an MRI equals about a billion pixels in our images,", Dijo.
Mapping the brain in such extraordinary detail normally involves a trade-off: thin sections of tissue can be imaged in very high resolution, but in thicker sections, the resolution falls. Kasthuri got around the problem [of having to manually photograph thousands of thin slices] by automating the process. Thin sections are imaged by the microscope and then stacked together.
“This work has really hit the sweet spot of extraordinary resolution, yet with the ability to survey relatively large anatomical regions,” said Arthur Toga, who uses brain imaging to study neurological disorders at the University of Southern California, Los Angeles.
“Other efforts have sacrificed one or the other. They provide high resolution in very small, often 2D, areas or they cover larger areas but without the detail provided here,", Dijo.
The machine will be used to take snapshots of the brain after death. The team hopes that by tracking neural pathways, they can answer questions about what a neurological disorder looks like in the brain. “If we could make a map of a brain with schizophrenia and compare it to one without schizophrenia, we can look for inappropriate connections that may contribute to the disorder,” Kasthuri said.
Sin embargo, Maria Ron, emeritus professor of neuropsychiatry at University College London, said it would be hard to use the system on large enough groups of patients and controls to reveal much about brain diseases.
The task would take extreme computing power too. An image of a whole mouse brain taken with the machine would create billions of gigabytes of data, Kasthuri said. That is equivalent to billions of high definition movies, and would create the largest dataset ever collected.
But Kasthuri is confident that computing power and storage will make this possible in three to five years. To image the human brain could another 10 año, he estimates.
The team is now working on machine vision which uses computers to track individual neurons through the brain and analyse their connections. “Once we have that in place, we have a potential showstopper,” Kasthuri said.
guardian.co.uk © Guardian News & Media Ltd. 2010