Share This Page
In 2010, a consortium of researchers began the Human Connectome Project, an ambitious undertaking to map the neural circuits of the living human brain. The neuroscience community has lauded the project as a game changer in our understanding of brain development and function. But the project’s approach, pooling neuroimaging data across a large number of participants, does not offer any insights into individual differences and how they might influence behavior and disease states. To make up for that lack, Russell Poldrack, now at Stanford University, started the My Connectome study, regularly scanning his own brain to see how its activity might change over the course of a year and how it might compare to the variability observed across a group. The initial results were published in the August 5, 2015 issue of Neuron; further results were published today in Nature Communications.
The quantified self
In 2012, Stanford geneticist Michael Snyder made headlines when he became the guinea pig for his own study tracking tens of thousands of proteomic and genomic markers over the course of 14 months. That work showed the value of following one’s personal health state at detailed resolution in both predicting and managing disease. Poldrack, then at the University of Texas at Austin, had been considering a similar study design himself—tracking the activity of a single human brain over time.
“One of the things we’ve known about psychiatric disorders for a long time is just how variable they are over time. If you look at the data on disorders like schizophrenia or depression, you can see changes in their brains over the course of just a few weeks. And we don’t really understand why,” he says. “My general philosophy has always been that to understand what’s happening in the brains of people who have illnesses we have to first understand how those relevant functions work in healthy brains. I had been thinking for some time that it might be interesting to follow an individual brain and see what kind of variability was there over time, that it might be possible to ask some really interesting questions about brain function and behavior in doing so. And Snyder’s study gave me that final push to start collecting data. And to collect it on myself.”
Thus, the My Connectome project (jokingly referred to as the “Russ-ome” by some lab members) was born. Over the course of the next eighteen months, Poldrack jumped into the fMRI scanner approximately twice a week to measure brain circuit activity, also logging information about his sleep, blood pressure, pulse rate, weight, body mass index, mood, and immune state.
The single and the many
Poldrack, in partnership with Steven Petersen, a cognitive neuroscientist at Washington University in St. Louis, ended up analyzing data from more than 80 imaging sessions. In their first publication for the project, they reported that Poldrack’s brain networks looked remarkably like averaged scans of grouped individuals. But they also discovered some interesting variations—particularly that Poldrack exhibited the most fluctuations of activity in visual and somato-motor regions of the brain.
“We didn’t really have a lot of hypotheses about what we would find because this really was discovery science. No one had ever looked at how the individual brain varies over this kind of length of time before,” he says. “The most surprising finding, to us, was the variability in what are fairly primary regions, the early visual regions and the somatosensory regions. Because, when you look across people, you primarily see differences between areas like the prefrontal cortex.”
Marcus Raichle, a neurologist at Washington University in St. Louis School of Medicine, says this study is a fascinating approach to better understand individual differences in brain activity.
“Way back in the 1980s, we toyed with the idea of doing image averaging, to get better signal-to-noise [measures] in our brain images. It was a big debate because we wondered if people might be so different that you couldn’t average,” he says. “It turned out that averaging was successful, and it’s the way we’ve been doing brain imaging studies since. But when you do those averages, you leave something on the cutting-room floor. That’s individual differences. And this study shows that there may be some really important data left behind that we should be looking more closely at.”
The brain and the body
In a second study, published the Dec. 9 issue of Nature Communications, Poldrack looked at correlations between his resting brain activity with gene expression, physiological state, immune function, and metabolomics, as well as behavioral data on his caffeine and food intake and his psoriasis symptoms. They found clear relationships between a variety of different measures, including Poldrack’s self-rated psoriasis severity as well as his eating habits.
“This was a really noisy scale. It was basically a self report of, ‘How flaky is my scalp from day to day?’” Poldrack says. “But we still saw strong relationships between my report and these different immune components. It’s interesting that even with such a noisy measure of disease severity, we can still pull out these relationships.”
But Poldrack was most surprised by how strongly related food was with metabolomic dynamics and gene expression. “We know that you are what you eat. But this made it clear that there are strong relationships between different aspects of biological function and the things I was eating,” he says.
Yet he is cautious about what these findings might mean. “These are all interesting relationships. But they are hard to interpret. Still, they can offer us new leads, particularly when we think about psychiatric disorders, about what might be going wrong in the brain and the body.”
Fewer participants, more time in the scanner
One of the most solid—and encouraging—findings from Poldrack’s project are how reliable current human functional connectivity measures are. Timothy Verstynen, a neuroscientist at Carnegie Mellon University, says that the beauty here is that when scans are repeated again and again, the results are highly consistent.
“What you see is that you can reliably capture the same patterns over and over and over again,” he says. “And that tells us that when we only take one scan, as we so often do, we can be fairly certain that what we see is actually what’s there and not just noise.”
But beyond reliability, Verstynen also says that doing more studies like these, or studies that follow a smaller number of participants over longer periods of time, can help us better understand individual differences and, ultimately, what may be going awry in the brains of people with psychiatric disorders. It may change the way that neuroscientists design and conduct neuroimaging studies in the future—reducing the number of participants and upping the number of scans for each.
“To study individual differences, we are seeing that we need to collect a lot of data, longer sessions of data, over time, in order to get reliable, statistically comparable results,” says Verstynen. “And we’re now seeing a push in neuroscience to do really careful, really basic levels of analysis so we can find the robustness we need to say things about individual variability and individual differences that you can’t really get in the sample sizes and over the session times we typically test in neuroimaging.”
Raichle agrees. “At first glance, having someone just scan their own brain may seem a little obsessive. But it’s fascinating. Because it finally gives us the ability to ask, ‘What does it take to really know Russ Poldrack’s brain? Or anyone else’s brain, for that matter?’” he says. “And it is going to take studies like these to track down the signs in something like depression, schizophrenia, or autism and help us understand what’s going wrong and then maybe use those differences to predict the future of their development. It’s an achievable challenge—but we have quite a bit of work ahead of us.”