Decoding the neural basis of resting-state functional connectivity mapping

Abstract Resting state functional magnetic resonance imaging (rs-fMRI) is an important modality for imaging the human brain. Capturing fluctuations in the blood oxygen level dependent (BOLD) signal while the brain is `at rest', rs- fMRI can detect distant, and often bilaterally-symmetric regions where activity is synchronized. Such regions are inferred to have `functional connectivity', and patterns of these networks have been found to be altered in a wide range of otherwise indistinguishable disease states. However, despite widespread use of rs-fMRI, interpretation of functional connectivity networks is limited by: 1) The dependence of fMRI BOLD signals on hemodynamic changes as a proxy for neural activity and: 2) A limited understanding of the mechanistic basis of functional connectivity networks in the context of cellular-level interactions and neural representations. In a recently published study, we demonstrated a new optical imaging technique capable of capturing both neural activity and hemodynamics across the bilaterally exposed superficial cortex of awake, behaving mice. This method revealed striking patterns of resting-state neural activity in the awake brain, exhibiting bilateral symmetry and features consistent with resting-state networks. Moreover, we demonstrated that this `neural network activity' was predictive of patterns of resting-state hemodynamics (via a linear convolution model), suggesting that we were visualizing the neural basis of resting state functional connectivity mapping. In the current proposal, we plan to leverage this new view of neural network activity in the brain, to characterize its cellular dependencies, pathways, drivers, behavioral correlates and interactions with hemodynamics. Data will be acquired using novel measurement and circuit manipulation techniques in awake, behaving mice, in addition to analysis of human rs-fMRI, intracranial and intraoperative electrophysiology in patients undergoing epilepsy evaluation (data from ongoing trials) as well as new intraoperative simultaneous optical hemodynamic and electrocorticography recordings. A major aspect of this project will be the aggregation of this data to generate predictive mechanistic and mathematical models of 1) Neural network activity and its dynamic properties and representations across scales and modalities and 2) The coupling relationships between resting-state activity in specific cell types and hemodynamics. These models will be used to derive and test improved methods for rs-fMRI acquisition, analysis and interpretation. To perform this work, we have assembled a world-class interdisciplinary team consisting of neuroscientists, neuroengineers, neurosurgeons, statisticians and experts in resting state fMRI acquisition and analysis. With a sharper understanding of the properties of neural network activity, its dependencies, and how best to harness it in human rs-fMRI, the results of this work could ultimately provide a mechanistic basis for network dysfunctions and their cognitive and behavioral manifestations in disease, potentially yielding new targets for therapies and more robust rs-fMRI based disease detection.