To become an independent and interdisciplinary computational neuroscientist, I have outlined a training plan in this proposal to: (a) develop broader knowledge of psychiatric and developmental disorders; (b) enhance theoretical and applied skills in advanced computational methods (including topological data analysis and graph theoretical modeling); and (c) refine skills in novel neuroimaging methods. I propose to develop and apply a novel computational framework to better understand the dynamical organization of intrinsic brain activity (IBA) in healthy participants and individuals with fragile x syndrome (FXS. To study the spatiotemporally rich phenomenon of IBA, resting-state functional Magnetic Resonance Imaging (rs-fMRI) data is typically analyzed by estimating statistical interdependence between time-varying signals from distinct brain regions over an entire scan period. By collapsing metrics over time, the resulting characterization only embodies an average snapshot across the complex phenomenon of IBA. Accordingly, there is a growing momentum towards quantifying the fluctuations in IBA. Several methods have been proposed, but they invariably average the data in either space (using seed- or network-based approaches) or time (using sliding- windows), thereby avoiding the examination of IBA in its entirety. I believ that comprehensive spatiotemporal analysis of IBA holds the key to finding person- and disorder-centric biomarkers. To this end, I first propose to develop methods that can examine dynamics of rs-fMRI data while preserving space and time information. Our preliminary results are promising, suggesting that using topological data analysis, developed at Stanford, we can mine high-dimensions of rs-fMRI data while addressing the crucial issues of low SNR, statistical confidence, validity, and reliability of the proposed methods. Second, I propose to use graph theory and state- space modeling to mathematically analyze and quantify the state/network transitions in IBA. Such modeling will not only improve our understanding of the mechanisms underlying dynamical brain organization in healthy and FXS groups, but will also allow us to generate concrete and testable hypotheses for future research.
For Aims 1 and 2, already collected rs-fMRI data in healthy and FXS groups will be used. We will also inspect the association of proposed metrics with behavioral and neuropsychological assessments to potentially identify novel disorder-centric biomarkers. Further, given the immense demand for resting-state biomarkers in pediatric and clinical populations for early detection of disorders, I propose to explore neuroimaging modalities that are both naturalistic and clinically relevant (esp., near-infrared spectroscopy (NIRS)). To extend the translational outcomes of my work, I also propose to port the methods developed for rs-fMRI (in Aims 1 and 2) to rs-NIRS platform and rigorously test the feasibility of rs-NIRS as an ancillary method of examining IBA in healthy adults and children. Altogether, I intend to reveal the dynamical structure of IBA in its entirety, thereby facilitating the discovery of person- and disorder-centric biomarkers in healthy participants and individuals with FXS.
Despite the burgeoning literature showing group-based differences in brain activity at rest, its specific association to any particular disorder is still lacking, making it an ineffective biomarker. We also lack applicable translation of group-based results to personalized clinical care. The proposed research will be particularly focused on developing individualized metrics to capture and computationally model the dynamics of brain activity at rest, so that specific biomarkers can be developed for early detection and treatment of mental illness.
| Green, Tamar; Saggar, Manish; Ishak, Alexandra et al. (2018) X-Chromosome Effects on Attention Networks: Insights from Imaging Resting-State Networks in Turner Syndrome. Cereb Cortex 28:3176-3183 |
