As the Human Connectome Project comes into its final year, Kamil Ugurbil, PhD, director of the Center for Magnetic Resonance Research (CMRR) at the University of Minnesota, and his team of researchers are working diligently to complete a dataset of advanced MRI that will undoubtedly affect the way we look at brain connectivity and function.
As much as the Human Connectome Project is about better understanding connectivity in the human brain through advanced MRI technology and techniques, it’s also about fostering a more open-access environment where researchers can share data across disciplines and locations.
Neurology Advisor sat down with Kamil Ugurbil to talk about high magnetic fields, the importance of a large subject pool, and what researchers can expect to get out of the growing dataset.
Neurology Advisor: The Human Connectome Project consists of two cooperative agreements, including a consortium of teams from the University of Minnesota, Washington University in St. Louis, and Oxford University. What role does each team play in the project?
Kamil Ugurbil, PhD: The University of Minnesota’s Center for MRI Research, which I direct, is responsible for developing the techniques for image acquisition and image reconstruction. We developed new methods of image acquisition and a unique 3 Tesla (T) instrument that we have now passed on to Washington University, and are continuing to conduct imaging with the 7T instrument at CMRR. Washington University is also responsible for data analysis and visualization, as well as data storage and distribution, which is no trivial task, especially with such a large database. Oxford is also very much an equal partner in this endeavor, and they are responsible for data analysis.
NA: CMRR has been instrumental in developing techniques and instruments with high magnetic fields, namely the 7T instrument being used for the project. How are the 3T and 7T instruments being used differently?
Ugurbil: We [CMRR] pioneered high magnetic fields, as we developed the 7T instrument. It’s now a commercial instrument, but for many years, it was just a kind of a home-built instrument at the University of Minnesota.
Higher magnetic fields will give you much more spatially accurate signals. In other words, what you detect and the activation in a 7T instrument has much better spatial correspondence to actual area of neuronal activities compared to a lower magnetic field. Plus, you gain signal-to-noise ratio that means you can get more accurate functional images at a higher spatial resolution.
In the HCP, 3T data resting state functional images are acquired with a spatial resolution of 2x2x2 mm3. This is a nominal resolution though, and the actual resolution is somewhat worse than that. 7T data is being acquired at 1.6×1.6×1.6 mm3, so volumetrically it is approximately a factor of 2 smaller. Again, this is nominal.
The real gains are somewhere between a factor of 2 to 3 in the three dimensional spatial resolution. The images have a much higher contrast and noise ratio, so we can detect many more networks. If we are able to identify approximately 100 networks, for example, that are unique networks from the functional imaging data at 3T, we can identify many more, approximately twice as much, using the 7T data. This is one of the major gains.
In the diffusion imaging data, the 7T data is higher resolution and has some gains over the 3T data in terms of connectivity. It loses out relative to the 3T data because of other reasons. We will have both 3T and 7T data on 200 of the 1,200 subjects in the project, and we plan on combining the 3T data, which has some gains for the diffusion weighted imaging, and the 7T data, which has different gains for diffusion imaging, in order to get the best of both worlds.
NA: Your subject pool for the project far exceeds the original task put out by the NIH. What are the advantages of having a larger subject pool, and why was the inclusion of twins and siblings integral to research?
Ugurbil: It was important to have a large subject pool of 1,200 people because we are looking at consequences of genes in populations. In order to really discern effects of genes vs. environment, for example, you will need a fairly large subject pool. We originally considered 2,000 subjects, but that was too ambitious based on the timeline, so we ended up with 1,200.
We collected data from twins and non-twin siblings because of the opportunity to ask questions about heritability of networks and environmental influences on networks.
NA: The research teams have already released a large amount of data, most recently in December. What do you hope to get out of the data being made public?
Ugurbil: These data are extremely complex and rich. You can take the same data and you can spend an incredible amount of time exploring it for different aspects. For instance, some people may be interested in the visual cortex and issues related to vision rather than the auditory cortex, or the relationship of auditory-visual pathways or decision making — you name it.
There are many different questions you can ask with these data, which means that these data have to be publicly available so that people can ask all these questions. If we just keep it to ourselves, we may be able to answer one or two questions, three maybe, in the lifetime of careers, but then the data will not be exploited as much as if it was publicly available.
This is discovery science, the science where you can actually have a huge number of investigators and have a huge number of different questions examined on a massive data set.
Once you have these datasets, people can also expect that the investigators or the research teams will improve the data analysis techniques, how they look at these data, and not only ask different questions, but also improve the techniques by which those questions can be asked or will be asked.
Kamil Ugurbil, PhD, is director, Center for Magnetic Resonance Research and holds the McKnight Presidential Endowed Chair Professorship in Radiology, Neurosciences and Medicine at the University of Minnesota. Ugurbil is co-principal investigator of the Human Connectome Project and is a member of the NIH BRAIN working group.