December 2016

A New Landscape: Brain Mapping With MRI
By Kathy Hardy
Radiology Today
Vol. 17 No. 12 P. 18

Think of a brain MRI in the same vein as a satellite image of the Earth's surface, and you'll have a clear picture of what neuroscientist David Van Essen, PhD, and a team of researchers are doing as part of a project to create a new map of the brain. Just as landscapes differ from place to place in the world—geographically, culturally, and politically—so too do the hills and valleys of the brain. Mapping is an important navigational tool for discovering locations on the globe and in the brain.

To meet this challenge, the National Institutes of Health–funded Human Connectome Project (HCP) was launched in 2010. The HCP is a consortium of institutions led by Washington University in St. Louis, the University of Minnesota, and Oxford University. Together, the group of researchers is navigating its way through the human brain and has studied 1,100 healthy young adults using noninvasive neuroimaging modalities. The goal in updating the brain's map is to see not only how it works but also to uncover information regarding brain connectivity, its relationship to behavior, and to see areas where certain disorders and disease may be observed. In addition, by successfully charting the human connectome in healthy adults, researchers hope to pave the way for future studies of brain activity during development and aging.

Form and Function
Van Essen, who is HCP's lead investigator; fellow neuroscientist Matthew Glasser, PhD, who is the lead author; and other researchers at Washington University in St. Louis and elsewhere used three different imaging technologies—conventional structural MRI, "resting-state" functional MRI (fMRI), and task-based fMRI—to gather data that enabled them to create an updated map of the human brain. Results of this study revealed evidence for 180 areas in the human cerebral cortex, 97 of which had not previously been identified. The researchers say high-definition brain scanning such as this may help identify biological markers for brain disorders such as schizophrenia and autism, as well as assist neurosurgeons in identifying critical areas to avoid when surgically resecting tumors in the brain.

"Each discrete area of the brain has a unique fingerprint," Glasser says, "meaning that anybody who [reads] MRI scans with a sufficient amount of high-quality data should be able to identify the same set of cortical areas."

When Glasser joined the HCP, his goal was to make better maps of the brain. Joining Van Essen in his work, the two soon learned that they needed better data and methods of analysis to create a 21st century map. With this work, they are mapping what has been poorly charted territory.

"Using a combination of structural and functional imaging modalities, we were able to improve the mapping techniques and ultimately generate a better map," Van Essen says. "This effort to simultaneously analyze three types of imaging data helps to achieve consistency in a discipline known for inconsistent results."

Part of what makes brain mapping difficult is the differences between neighboring cortical areas, which are not always evident when using only a single imaging modality, making it important to concurrently examine structure, function, and/or connectivity. For example, the folds of the cerebral cortex allow a large surface area to fit inside the skull, but they vary widely among individuals; Van Essen describes them as analogous to a newspaper that is crumpled to fit inside a skull-sized box.

"Looking at the cerebral cortex, which is the outer layer of the cerebrum, you see folded gray matter. Each person has a distinct and highly complex folding pattern," Van Essen says. "You can't align brains accurately just by relying on the pattern of folds."

The HCP researchers made use of measures that included not just cortical topography but also cortical thickness, brain function, connectivity between regions, the amount of myelin—the insulation around axons that speeds up neural signaling—and neural activity. In addition, the investigators applied novel methods developed by HCP colleagues at Oxford University to align the cortical surfaces of different individuals using features based on where cortical areas are located, not just on the folding patterns. This dramatically improved the quality of intersubject alignment.

New Cartographic Tool
To take these measurements, the researchers used the MAGNETOM Connectome MR, an MRI research scanner from Siemens Healthineers built specifically for the HCP. According to David Carpenter, PhD, manager of emerging technologies for Siemens Healthineers, the company was approached by representatives of the HCP, who were looking for technology powerful enough to map the brain.

One important improvement to the MRI acquisition was the ability to acquire the brain images faster with simultaneous multislice (SMS) imaging, a technique that uses parallel image reconstruction to accelerate the data acquisition time. This allows high resolution images to be taken more quickly, a particularly important consideration with fMRI.

"We're able to acquire images more rapidly, and with speed you get more data," Glasser says. "Speed also helps to separate the signals and clear up the data."
"With SMS, imagers can complete a comprehensive brain map in about 10 minutes, as opposed to an hour," Carpenter says.

Work conducted within the scope of the HCP had an impact on MRI advances within clinical radiology. As Carpenter explains, the current Siemens Healthineers MAGNETOM Prisma clinical system evolved from the MAGNETOM Connectome device created for the HCP and brings to market almost double the gradient strength of current 3T systems; Carpenter adds that the MAGNETOM Prisma, along with the work being done by the HCP "brings research power into the clinical world."

"There are only two MR devices like this in the world, with a gradient coil of this strength," Carpenter says. "Each MRI imaging type used in the brain mapping process relies on the strength of the gradient coil. The system has applications that display functional information and enable the visualization of details in very small anatomical structures, such as subtle details found in the brain."

Van Essen notes that the increased gradient strength of the MR system version used by the HCP only benefits diffusion MRI. For technical reasons, the high-quality diffusion imaging acquired by the HCP was not used for the brain parcellation. He adds, however, that the HCP's improved brain parcellation will greatly benefit analyses of "structural connectivity"—the charting of trajectories of fiber bundles that travel through the brain's white matter—using diffusion imaging data.

Carpenter says that as neurosurgeons have learned of the work being done with the MAGNETOM Connectome MR system, they have shown interest in incorporating the device's brain-mapping capabilities into their surgical planning.

"The neuroradiology community is excited to see further developments in this area," he says.

21st Century Map 1.0
An update to the human cerebral cortex map is long overdue. The cortical map used most often today by researchers and clinicians is essentially the same one developed more than 100 years ago by anatomist Korbinian Brodmann. The German scientist delineated approximately 50 basic cortical regions.

Van Essen started mapping the cerebral cortex in the 1970s.

"We knew it was important to determine the most accurate number of areas of the brain, but, more importantly, we needed to find out where the distinct areas of the brain are located and the functions associated with each area," Van Essen says. "The single resting state is helpful but does not provide a complete view of all regions."

Earlier maps are based on unimodal mapping, he explains, which is a single type of measurement. This method of mapping provided a limited view of the brain. The new map, being based on multiple MRI measurements, provides greater detail.

Glasser says he looked for areas in the cerebral cortex where he saw significant changes in two or more properties and used them to delineate borders on the map. With functional imaging, for example, Glasser says that subjects' brains were imaged with MRI while at rest, then while involved in tasks such as listening to stories or doing math. In each case, researchers looked at images to see which areas of the brain were active, as well as which areas showed deactivation. In the end, they combined the data from all aspects of imaging.

"Also, we looked to see how correlated the MRI signal was over time between different brain regions when the subject was at rest, a measure of 'functional connectivity,'" Glasser says. "Measurements were changing. We defined areal borders in locations where multiple independent measurements showed change to make our new map of the cerebral cortex."

Van Essen describes one new area, 55b, which is located in the frontal cortex. This is a small area of the brain that has been mostly overlooked, appearing as a blur in previous maps. However, it plays a role as part of the brain's language network, including speech production.

Van Essen's and Glasser's group developed an algorithm that will now make their data collection protocol available to other researchers and clinicians. With research continuing among the HCP scientists, both Van Essen and Glasser refer to this version of the updated brain map as "version 1.0."

"This isn't the final map, but a better map than the version clinicians and researchers were working with," Van Essen says.

— Kathy Hardy is a freelance writer based in Phoenixville, Pennsylvania. She is a frequent contributor to Radiology Today.