Considerable efforts have been spent in the past two decades to search for biomarkers for pre-symptomatic stages of Alzheimer's disease (AD). For neuroimaging, amyloid-PET imaging of amyloid beta (A?) accumulation in the brain is considered an early marker for the preclinical stage of AD, while tau-PET imaging correlates more closely with neuronal injury and cognitive decline. However, PET scans are expensive and involve radioactive tracers. Resting state fMRI (rs-fMRI) studies in AD have shown that the functional connectivity (FC) of resting brain networks is progressively diminished in subjects with mild cognitive impairment (MCI) and AD. However, FC analysis of rs-fMRI has limited capability to characterize the dynamic fluctuations of rs-fMRI signals that possess clinically meaningful information. Our group and others have recently explored the use of entropy measures as indices of the complexity and regularity of rs-fMRI time-series. Accumulating data showed decreasing entropy values associated with aging, APOE ?4 genotype, cognitive decline in autosomal dominant Alzheimer's disease (ADAD) and late-onset AD (LOAD). Our group began developing the Complexity Toolbox in 2013 as the first systematic and comprehensive software package dedicated to complexity analysis of neuroimaging (fMRI) data. In particular, a recent independent study using our toolbox to analyze the rs-fMRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study reported progressive reductions of entropy from healthy controls, early MRI, to late MCI and AD groups, with significant associations between complexity measures of rs-fMRI and cognitive decline in MCI/AD subjects. Our preliminary data in ADAD and LOAD subjects further showed consistent negative correlations between rs-fMRI entropy and tau-PET signal. The goal of this project is to further develop our Complexity Toolbox and a cloud-based pipeline for comprehensive complexity analysis of (large scale) fMRI data. We will systematically evaluate the complexity of fMRI as a novel imaging marker of AD in both ADAD and LOAD populations, using 3 public databases of rs- fMRI and PET including Dominantly Inherited Alzheimer Network (DIAN), Connectome of ADAD, and the Alzheimer's Disease Neuroimaging Initiative (ADNI-3) with a total sample size >900. Finally, we will use advanced machine learning techniques to evaluate complexity of rs-fMRI as a predictor for transversion from healthy to MCI and to AD. We will generate a disease staging model based on multimodal AD biomarkers including PET, CSF and rs-fMRI measures. We hypothesize that the complexity of BOLD signals provides an index of the information processing capacity of regional neuron populations, and is therefore sensitive to tau- related neuronal injury and cognitive decline in the AD processes. The successful completion of this project will lead to a noninvasive, economical and alternative imaging biomarker of neuronal injury in MCI and AD with relevant tools ready to be deployed in clinical research and care of AD.
Biomarkers for pre-clinical stages of Alzheimer's disease (AD) have become increasingly important for the development of preventative interventions. For neuroimaging, positron emission tomography (PET) imaging of amyloid beta (A?) and tau protein provide early imaging markers of AD and can track disease progression but are expensive and require the use of radioactive tracers. This project aims to develop and evaluate a noninvasive, quantitative and economical imaging marker of AD based on the complexity or regularity of resting state functional magnetic resonance imaging (fMRI).
