Despite its modest physical profile — approximately three pounds of water and fatty tissue — the brain is the most important and complex organ in the human body. The human brain regulates physiological processes while concurrently allowing us to interpret and interact with the world around us.
Advances in imaging technology over the last century have ushered critical insights into the brain’s anatomy and function. Current research efforts seek new methods to robustly estimate associations in the functional activity of spatially distinct brain regions via fluctuations in the blood oxygen level dependent signal (BOLD) in fMRI or perturbations in cortical electrical signals measured with EEG. Although the list of measurements and technologies for acquiring signals are becoming increasingly sophisticated, various artifacts, such as cardiac and respiratory rates as well as data preprocessing artifacts, make it difficult to separate functionally relevant signals from noise. Unfortunately, this can lead to erroneous functional connectivity findings.
There is obviously a need to utilize auxiliary information to more accurately estimate functional connectivity. One particular avenue under heavy investigation over the last fifteen years is to leverage the brain’s structure in studies of functional connectivity. In fact, there is a strong biological basis for this line of reasoning. The brain’s fundamental functional processing unit is the neuron, which is a cell body responsible for receiving, interpreting, and transmitting signals to other neurons. These cells are constantly active, coordinating signals that produce a wide spectrum of cognitive and physical actions. It is a well-known fact that neurons transmit their signals via fibers wrapped in fatty tissue. En masse, these fiber cables serve as a telecommunication network in the brain, permitting fast and efficient routing of messages between neurons. Clearly, neurons that “wire together, fire together.”
Logically, mapping the brain’s telecommunication network is simple: enumerate all fibers connecting each pair of neurons. In the average adult, there are approximately 100 billion neurons, making this simple task computationally daunting and infeasible for statistical analyses. Instead, researchers group thousands of neurons into a volumetric unit called the voxel. Spatially proximal voxels are subsequently grouped to represent a region of interest (ROI). At this level, communication between billions of neurons in the adult human brain can be summarized by 100-1000 ROIs.
Even at this coarser scale of analysis, various researchers have shown that there is a strong, albeit nebulous, relationship between structure and function in the brain. Although counterintuitive, the brains of subjects who do not engage any specific cognitive task organize into tightly knit functional blocks whereby the regions within the module exhibit strong co-activity. It has been hypothesized that the brain maintains this coordinated spontaneous functional activity in order to quickly respond to external stimuli as well as regulate internal emotions and thoughts. This resting-state task-free design is widely used and has led to important discoveries about brain function in disease (Greicius et al., 2007; Lynall et al., 2014).
Others have found strong relationships between function and structure in the brain at rest. Kemmer et al. (2017) show that functional modules with robust functional connections are typically composed of ROIs exhibiting strong structural connections. Honey et al. (2009, 2010) show that strong direct structural connections are not always a necessary precursor for strong functional co-activity. Skudlarski et. al (2008) show that indirect structural connectivity, i.e. connections between two regions via intermediate ROIs, potentially explain oddities in the function-structure relationship.
Many methods try to incorporate brain structure into the estimation of functional connectivity (Hinne et al., 2014; Ng et al., 2012; Pineda-Pardo et al., 2014; Xue et al., 2015). Our method, which is described in the article entitled Integrative Bayesian analysis of brain functional networks incorporating anatomical knowledge published in the journal NeuroImage, similarly utilizes brain structure in constraining functional connectivity estimation.
Unlike its predecessors, our method flexibly incorporates structural connectivity information into the estimation routine, permitting strong functional connections supported by the rs-fMRI data that do not exhibit strong structural connections. Furthermore, the existence of a functional connections is not constrained by the presence or absence of a structural connection. Additionally, our method is more robust to inaccurate structural connectivity estimates than competitors. This is especially important given the improving, yet imperfect, technologies for estimating structural pathways in the brain. Finally, our proposed approach produces reliable estimates of functional connectivity across scanning sessions. It is well known that that brain imaging data exhibits large variability within subjects, during and across scanning sessions. Clearly, our approach disentangles, to some degree, functionally relevant signals from those introduced by noise artifacts by leveraging knowledge of structural pathways in the brain.
There are also clinical implications for the proposed methodology. As discussed in the article, one can study the impact of structural connections on brain functional development as individuals age. Furthermore, our approach could provide insights into functional connectivity in diseases that diminish the integrity of the brain’s white matter pathways. There are various practical and clinically meaningful reasons to incorporate auxiliary imaging information when available. Critically, our understanding of the brain in health and disease hinges upon the ability of the research community to flexibly model fundamentally different, yet complimentary sources of information.
These findings are described in the article entitled Integrative Bayesian analysis of brain functional networks incorporating anatomical knowledge, recently published in the journal NeuroImage. This work was conducted by Ixavier A. Higgins, Suprateek Kundu, and Ying Guo of Emory University’s Department of Biostatistics and Bioinformatics. The work was supported by the National Institute of Mental Health of the National Institutes of Health under Award Number ROI MH105561 and R01MH079448.
- Greicius, M. D., Flores, B. H., Menon, V., Glover, G. H., Solvason, H. B., Kenna, H., … & Schatzberg, A. F. (2007). Resting-state functional connectivity in major depression: abnormally increased contributions from subgenual cingulate cortex and thalamus. Biological psychiatry, 62(5), 429-437.
- Hinne, M., Ambrogioni, L., Janssen, R. J., Heskes, T. and van Gerven, M. A. (2014), ‘Structurally-informed bayesian functional connectivity analysis’, NeuroImage 86, 294– 305.
- Honey, C. J., Thivierge, J.-P. and Sporns, O. (2010), ‘Can structure predict function in the human brain?’, Neuroimage 52(3), 766–776.
- Honey, C., Sporns, O., Cammoun, L., Gigandet, X., Thiran, J.-P., Meuli, R. and Hagmann, P. (2009), ‘Predicting human resting-state functional connectivity from structural connec- tivity’, Proceedings of the National Academy of Sciences 106(6), 2035–2040.
- Kemmer, P. B., Bowman, F. D., Mayberg, H. and Guo, Y. (2017), ‘Quantifying the strength of structural connectivity underlying functional brain networks’, arXiv preprint arXiv:1703.04056 .
- Lynall, M. E., Bassett, D. S., Kerwin, R., McKenna, P. J., Kitzbichler, M., Muller, U., & Bullmore, E. (2010). Functional connectivity and brain networks in schizophrenia. Journal of Neuroscience, 30(28), 9477-9487.
- Ng, B., Varoquaux, G., Poline, J.-B. and Thirion, B. (2012), ‘A novel sparse graphical approach for multimodal brain connectivity inference’, Medical Image Computing and Computer-Assisted Intervention–MICCAI 2012 pp. 707–714.
- Pineda-Pardo, J. A., Bruña, R., Woolrich, M., Marcos, A., Nobre, A. C., Maestú, F. and Vidaurre, D. (2014), ‘Guiding functional connectivity estimation by structural connectivity in meg: an application to discrimination of conditions of mild cognitive impairment’, Neuroimage 101, 765–777.
- Skudlarski, P., Jagannathan, K., Calhoun, V. D., Hampson, M., Skudlarska, B. A. and Pearlson, G. (2008), ‘Measuring brain connectivity: diffusion tensor imaging validates resting state temporal correlations’, Neuroimage 43(3), 554–561.
- Xue, W., Bowman, F. D., Pileggi, A. V. and Mayer, A. R. (2015), ‘A multimodal approach for determining brain networks by jointly modeling functional and structural connectivity’, Frontiers in computational neuroscience 9.