Neuroscience is going high-def.
Researchers have developed a series of technological improvements that dramatically increase the spatial resolution of an fMRI (functional magnetic resonance imaging) machine. The improvements allow an fMRI to image voxels—the 3D equivalent of pixels—that are less than half a millimeter on each side. In doing so, they have passed an important threshold relative to the structure of the human brain, to roughly match the scale of functional clusters of neurons.
The NexGen 7T scanner is a ten-fold improvement in resolution over 7T MRI machines that are currently available to researchers. The finer scale could be useful for many research questions. For example, higher resolution images could provide answers about the broad functional organization of some brain areas and the activity of neurons in different layers of other areas.
The new scanner is “an exciting step forward,” said Serge Dumoulin, a neuroscientist and director of the Spinoza Centre for Neuroimaging in the Netherlands, who was not involved in development.
The overall effect is a spatial resolution more than 100 times as fine as widely-available fMRI, and an order of magnitude beyond the detail of other 7T fMRI.
The international team, which included engineers at Siemens, described the new scanner in Nature Methods on 27 November. The work was funded in part by the University of California, Berkeley and the BRAIN Initiative from the National Institutes of Health.
A 7T MRI scanner, as the name suggests, generates powerful magnetic fields at 7 tesla. The NexGen machine keeps the main magnet and replaces most of the other major components. Rather than highlight any single component, the researchers say that multiple hardware improvements contributed to the higher resolution of the new scanner.
“We knew that if we improved the performance of major systems, we would get cumulative improvements, and each subsystem affected a different important parameter,” says co-author David Feinberg, a neuroscientist at UC Berkeley and a company called Advanced MRI Technologies.
Any fMRI scanner works by creating magnetic fields that affect the orientation of molecules in the brain. Once one magnetic field is set to a predictable pattern, densities or types of tissue can be differentiated by applying a second magnetic field and rapidly oscillating the orientation. Various techniques have been developed which use blood flow or oxygen levels as a proxy for brain activity.
The NexGen 7T includes a new gradient coil, which sets the pattern of the underlying magnetic field. The coil had perhaps the most novel design of the new elements, with an additional third layer of winding wires, and a redesigned cooling system. The transmit system, which controls the secondary magnetic field, and radiofrequency receiver array were both expanded to at least double the channels one would typically encounter in research MRI scanners, with 16 and 128 channels, respectively. Those changes created new challenges related to data acquisition, transmission, and encoding.
“When you have 128 receivers, the data size increases exponentially,” said co-author An T. Vu, a radiologist at the University of California, San Francisco. “The data storage and computational resources need to catch up, essentially.”
The overall effect is a spatial resolution more than 100 times as fine as widely-available fMRI, and an order of magnitude beyond the detail of other 7T fMRI. The voxel size has shrunk from a volume of about a microliter (a cubic millimeter) to smaller than 0.1 microliters under some circumstances, comparable to coarse grains of sand. On the scale of the brain and its cells, the scanner is still in an intermediate zone of resolution.
“We knew that if we improved the performance of major systems, we would get cumulative improvements, and each subsystem affected a different important parameter.” —David Feinberg, UC Berkeley
But with higher, sub-microliter resolution, neuroscientists should get a detailed look at the functional organization of groups of neurons in the cerebral cortex—the lobed upper areas of the brain—especially in layers or columns, said Dumoulin. It could also help reveal the detailed structure of other brain regions, such as the brainstem and cerebellum, which are “often beyond the field of view of conventional neuroimaging.”
The team chose to work with a 7T MRI scanner in part to strike a balance between technical requirements and availability compared to even more powerful, experimental MRI machines. Stronger magnetic fields can increase the potential for high-clarity images, but also increase the potential dangers of RF heating for people inside the scanner. So while only one NexGen scanner has been built so far (at UC Berkeley), any of the 100 or so operational 7T MRI scanners globally could in theory be retrofitted to the same specifications.
Higher resolutions could be useful beyond fMRI with other MRI research applications, for example, related to functional connectivity and brain metabolism. The researchers also had eventual medical uses in mind. “It’s designed for basic research, but as soon as there’s something promising and useful, we can immediately translate it to clinical experiments,” said Feinberg.
