To better understand the brain, look at the bigger picture

Summary: Zooming out to visualize larger areas of the brain using fMRI technology allows scientists to capture additional relevant information, offering a better understanding of neural interactions.

Source: Yale

Scientists have learned a lot about the human brain through functional magnetic resonance imaging (fMRI), a technique that can provide insight into how the brain works. But common fMRI methods may not contain key information and provide only part of the image, say Yale researchers.

In a new study, they assessed different approaches and found that zooming out and occupying a wider field of view captures additional relevant information that is overlooked and narrowed down, offering a better understanding of neural interactions.

Moreover, these more comprehensive results could help address the problem of reproducibility of neuroimaging, where some of the results presented in the studies cannot be replicated by other researchers.

The results were published on August 4 in Materials of the National Academy of Sciences.

Research that uses fMRI usually focuses on small areas of the brain. As one example of this approach, researchers are looking for areas of the brain that become more “active” when a specific activity is performed, targeting the smallest areas that are most activated. But growing evidence shows that brain processes, and complex processes in particular, are not limited to small parts of the brain.

“The brain is a network. It’s complicated, ‘said Dustin Scheinost, associate professor of radiology and biomedical imaging and senior author of the study. He said oversimplification leads to imprecise conclusions.

“For more sophisticated cognitive processes, many areas of the brain are unlikely to be completely disengaged,” added Stephanie Noble, a postdoctoral researcher at the Scheinosta lab at the Yale School of Medicine and lead author of the study.

Focusing on small areas ignores other regions that may be involved in the behavior or process being studied, which may also influence the direction of future research.

“You are creating this incorrect picture of what is actually going on in the brain,” she said.

As part of the study, researchers assessed how well fMRI analyzes at different scales were able to detect the effects or changes in fMRI signals when participants perform different activities, revealing which parts of the brain are involved.

They used data from the Human Connectome Project, which collected brain scans of people performing various tasks related to complex processes such as emotions, language and social interactions.

The research team looked for effects in very small parts of the brain’s network – such as connections between just two areas – as well as clusters of connections, vast networks and entire brains.

They found that the larger the scale, the better they were able to detect the effects. This ability to detect effects is known as “power”.

“These methods give us more power on a larger scale,” said Noble.

At the smallest scales, scientists were only able to detect about 10% of the effects. But at the network level, they could detect over 80% of them.

The trade-off for the extra power was that wider views did not convey information that was as spatially accurate as with smaller-scale analyzes. For example, at the smallest scale, the scientists could confidently say that the effects they observed occurred in a small area.

However, at the network level, they could only say that the effects were occurring on a large part of the network, not exactly where on the network.

The goal, says Noble, is to balance the advantages and disadvantages of different methods.

“Would you rather be very sure about a small fraction of the relevant information – in other words, have a very clear picture of only the tip of the iceberg?” she said.

“Or maybe you’d rather have a really big picture of the whole iceberg, which is maybe a bit blurry, but gives a sense of complexity and a wide spatial scale of where things are happening in the brain?”

For other researchers, this approach is easy to implement, and Noble said she was excited to see how other scientists are using it.

This shows a brain made of gears
Moreover, these more comprehensive results could help address the problem of reproducibility of neuroimaging, where some of the results presented in the studies cannot be replicated by other researchers. The image is in the public domain

Notes that the fields of psychology and neuroscience, including neuroimaging, face the problem of reproducibility. The low power of fMRI analyzes contributes to this: low power studies reveal only small pieces of history that can be seen as contradictory, not as parts of the whole.

Increasing the power of fMRI, as she and her colleagues have done here by scaling up their analyzes, may be one way to address reproducibility challenges by revealing how seemingly contradictory results can actually be harmonious

“Moving up the food chain, so to speak, going from a very low level to more complex webs is a lot more powerful,” said Scheinost. “This is one of the tools we can use to solve the reproducibility problem.”

See also

This shows the man playing the banjo

And scientists shouldn’t throw the baby out with the bathwater, said Noble. Much good work has been done to improve methods and increase stringency, and fMRI is still a valuable tool, she said: “I think judging power, discipline and repeatability is healthy in any area. Especially one that deals with the complexity of living entities and mental processes. “

Noble is currently developing a ‘power calculator’ for fMRI to help others design research in a way that achieves the desired power level.

About this news from neuroimaging research

Author: Mallory Locklear
Source: Yale
Contact: Mallory Locklear – Yale
Image: The image is in the public domain

Original research: Open Access.
“Improving Power in Functional Magnetic Resonance Imaging by Going Beyond Cluster Level Inference” by Stephanie Noble et al. PNAS


Improve the power of functional MRI by going beyond cluster-level inference

Inference in neuroimaging usually occurs at the level of focal areas of the brain or circuitry. Increasingly, however, well-conducted studies show a much richer picture of large-scale effects distributed throughout the brain, suggesting that many focal reports may only reflect the tip of the underlying effects iceberg.

How the focal or broad-scale perspective influences our conclusions has not yet been comprehensively assessed using real data.

Here we compare the sensitivity and specificity in procedures representing multiple levels of inference using an empirical comparative procedure that re-samples the task connectomes from the Human Connectome Project dataset (1000 subjects, 7 tasks, 3 re-sampling group sizes, 7 inference procedures).

Only the large scale procedures (network and whole brain) achieved the traditional 80% statistical power level to detect the mean effect, reflecting> 20% more statistical power than the focal (edge ​​and cluster) procedures. The power also increased significantly for the false-discovery rate – compared to the family’s error rate control procedures.

The downsides are quite limited; The loss of specificity in the large-scale and FDR procedures was relatively small compared to the increase in potency. Moreover, the large-scale methods we introduce are simple, quick and easy to use, providing a simple starting point for researchers.

It also points to the promise of creating more sophisticated methods for large-scale not only functional connectivity but also related fields, including task-based activation.

Taken together, this work demonstrates that re-scaling inference and the choice of FDR controls are both immediately achievable and can help solve the statistical power problems plaguing typical research in the field.

Leave a Comment