Liang Zhan - US grants

Alzheimer?s disease (AD) is the most common form of dementia, with the number of affected Americans expected to reach 13.4 million by the year 2050. While it is well known that AD leads to progressive neuronal death, the exact mechanism of AD remains elusive. Currently, a definitive diagnosis can only be reached by autopsy or brain biopsy, and the neurodegenerative processes in two AD patients can follow very different courses. Further, treatment options for AD remain limited, let alone cure. For this reason, non-invasive neuroimaging has been extensively investigated in the hope that it may provide more sensitive markers for screening and early detection of AD. Yet, despite the amount of resources devoted to AD imaging research, CSF Tau and A?42 continue to outperform any non-invasive imaging markers. Multimodal connectomics, including functional and structural connectome (derived from fMRI and diffusion MRI respectively), has the potential to gain system-level structure- function insights into the mechanisms of AD and thus offers a novel platform for developing new diagnostic strategies. Despite a number of interesting connectome findings in recent years, few of these connectome results have been replicated independently or proven clinically relevant, which can be partially explained by the sensitivity to parameter settings during preprocessing and connectome construction, such as the choice of brain parcellation and the type of fMRI time series correlations (full versus partial) or tractography (deterministic or probabilistic). Moreover, conventional connectome approaches usually focus on scalar summary statistics (e.g., nodal or edge-wise measures) using linear statistical techniques, which fit at each node (or edge) independent of other nodes (or edges) and thus discard important informative graph structure. Instead, this proposal will develop a multi-view connectome framework that homogenizes multiple instances of stable and reproducible high-level connectome properties across modalities and across spatiotemporal scales. This framework will be applied and cross-validated using two independent AD cohorts (Alzheimer's Disease Neuroimaging Initiative or ADNI and Wisconsin Alzheimer?s Disease Research Center cohort or Wisconsin ADRC). The identified connectome features can serve as the potential non-invasive markers for guiding the AD diagnosis.

Recent advances in multimodal brain imaging and high throughput genotyping and sequencing techniques provide exciting new opportunities to ultimately improve our understanding of brain structure and neural dynamics, their genetic architecture, and their influences on cognition and behavior. However, data privacy and security issues have inhibited data sharing across institutes. Emerging multi-site collaborative data analysis can address these issues and facilitate data and computing resource sharing. In collaborative data analysis, the participating institutes keep their own data, which are analyzed and computed locally, and only share the computed results by communicating with a server. The server communicates with all institutes and updates the local models such that the trained machine learning models indirectly use all data and are shared with all institutes. Although some distributed/parallel computation techniques were recently proposed to address big data mining problems, most of them are synchronous models. Asynchronous distributed learning methods are much more efficient, because they allow the server to update the model with information from only one worker node without waiting for slow worker nodes in each round. However, the convergence analysis for the asynchronous distributed algorithms is much more difficult due to the inconsistent variables update across nodes. Thus, it is challenging to design efficient distributed machine learning algorithms for collaborative big data analysis. The research objective of this project is to address the computational challenges in the emerging multi-site collaborative data mining for brain big data.

This project seeks to harness the opportunities of designing new efficient asynchronous distributed machine learning algorithms with rigorous theoretical foundations for multi-site collaborative brain big data mining, creating large-scale computational strategies and effective software tools to reveal sophisticated relationships among heterogeneous brain data. This project designs the asynchronous distributed machine learning and principled big data mining models to conduct the comprehensive study of brain imaging genomics and connectomics. Specifically, the principal investigators investigate: 1) collaborative genotype and phenotype association study using new asynchronous doubly stochastic proximal gradient algorithms; 2) communication-efficient multi-site collaborative data integration models to integrate imaging genomics data for predicting outcomes of interest; 3) collaborative deep learning algorithm speedup by the asynchronous distributed algorithms with applications in temporal cognitive change prediction; and 4) new graph convolutional deep learning models for brain network mining. It is innovative to integrate new distributed machine learning and data-intensive computing with brain imaging genomics and connectomics that hold great promise for a systems biology of the brain. The developed methods and tools impact other neuroimaging, genomics, and neuroscience research, and enable investigators working on brain science to effectively test their scientific hypotheses. This project will also facilitate the development of novel educational tools.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Synaptic dysfunction has been hypothesized to be one of the earliest brain changes in Alzheimer?s disease (AD), leading to hyper-excitation in neuronal circuits. However, network changes related to age and sex tend to overlap with disease neuropathology, increasing the difficulty of separating disease-specific alterations from those related to normal aging trajectories in males and females. Indeed, AD disproportionately affects women, who comprise two thirds of all persons diagnosed with AD dementia. Leveraging resting state fMRI connectome and diffusion MRI-derived structural connectome, we will use a novel hybrid resting-state structural connectome (rs-SC) to study excitation-inhibition balance. Recently, using a group of cognitively normal APOE-?4 carriers and age/gender matched non-carriers we demonstrated a sex-by-age-by-phenotype interaction, with significant hyperexcitation with increasing age only observable in women, but not in men. Further, hyperexcitation in female carriers began to exhibit at age 50 in the anterior cingulate, parahippocampal gyrus and temporal lobe regions, and the degree of hyperexcitation is linked to compensatory recruitment of neuronal resources during a spatial learning memory task. In this proposal, we will characterize 1) sex-specific normative trajectories of excitation-inhibition balance using the Human Connectome Project (HCP) data, and 2) altered excitation-inhibition balance in abnormal aging using the Alzheimer?s Disease Neuroimaging Initiative (ADNI) data, as well as 3) further test and validate our hyperexcitation framework in longitudinal mouse models of AD. RELEVANCE (See instructions): In this proposal, we will develop novel computational tools to characterize hyper-excitation patterns in aging and Alzheimer's Disease and validate our hyperexcitation framework on human data (ADNI and HCP) as well as longitudinal mouse models of AD. This will significantly improve our understanding of AD and potentially accelerate the discovery of more robust non-invasive imaging biomarkers of AD.