This project aims to further the development of diffusion MRI as a neuroimaging technology capable of identifying how complex patterns of neural connectivity develop in the brain, underlie cognitive functioning, and are modified by neurological disorders. It contributes directly to one of the goals listed in the NIBIB Strategic Plan: Develop innovative biomedical technologies that integrate engineering with the physical and life sciences to solve complex problems and improve health. Currently, a wide array of ad hoc computational methods are applied to diffusion MRI data, leading to research findings that are confusing, inconclusive, discordant between labs, or non-reproducible. The current project advances diffusion MRI toward more rigorous and standardized measurements of neural connectivity that are built from strong statistical theory. It will develop a multi-scale representation of brain structural connectivity which provides convenient and biologically interpretable multi-scale features that can be related to external variables of interests in statistical models. The immediate result of the project could be brain connectivity estimators whose performance characteristics are better understood and therefore generalize better across studies and laboratories. The longer-term result could be accelerated development of diagnostic tests, models of brain disease, and fundamental neuroscience knowledge, all derived from advanced diffusion MRI tools.
Specific aims i nclude: 1. Use a novel needlets based representation to provide a concise summary of local fiber orientations; 2. Construct an innovative multi-scale fiber orientation representation that aggregates direction information over variable-sized spatial neighborhoods, which makes it possible to determine the spatial scale of analysis in a data driven way; 3. Develop high-dimensional statistical models that assess multi-scale structural connectivity change over time and relation to external variables. These methods will then be validated through analytical studies as well as using both synthetic and real world data sets.
Understanding how brain network development in children is influenced by genetics, environmental exposures and neurodevelopmental syndromes, could have tremendous implications for policies to promote early-life brain health. Understanding how neural networks become disconnected over time with diseases such as Alzheimer's disease could have major implications for efforts to halt such damage.
Yan, Hao; Carmichael, Owen; Paul, Debashis et al. (2018) Estimating fiber orientation distribution from diffusion MRI with spherical needlets. Med Image Anal 46:57-72 |
Johnstone, Iain M; Paul, Debashis (2018) PCA in High Dimensions: An orientation. Proc IEEE Inst Electr Electron Eng 106:1277-1292 |
Zhou, Siyuan; Paul, Debashis; Peng, Jie (2018) Modeling subject-specific nonautonomous dynamics. Stat Sin 28:423-447 |
Wong, Raymond K W; Lee, Thomas C M; Paul, Debashis et al. (2016) FIBER DIRECTION ESTIMATION, SMOOTHING AND TRACKING IN DIFFUSION MRI. Ann Appl Stat 10:1137-1156 |
Wong, Raymond K W; Lee, Thomas C M; Paul, Debashis et al. (2016) REJOINDER: ""FIBER DIRECTION ESTIMATION, SMOOTHING AND TRACKING IN DIFFUSION MRI"". Ann Appl Stat 10:1166-1169 |