Diffusion MRI (dMRI) plays critical roles in understanding the neural underpinnings in the development of Alzheimer's disease (AD). Obtaining high quality image with high spatial and q-space resolutions requires prolonged data acquisition, which leads increased mental stress and higher risk of motion artifacts. Acquisition time is a particularly important issue in clinical dMRI in AD and at-risk subjects. But shortening data acquisition time through reducing image resolution will reduce the ability to estimate white matter tracts and degenerate the statistical power. In-plane acceleration (IPA), and simultaneous multislice (SMS), are potential solutions to balance imaging quality and acquisition time. However, IPA and SMS have important limitations, as they require reconstruction methods that introduce noise. The literature studying the impacts of acceleration on dMRI and dMRI-based structural connectome analysis is very limited. Some important questions remain open, including how IPA and SMS affect structural connectome estimation and whether we can further shorten the acquisition time without sacrificing image quality. In this study, we will collect and analyze repeated scans from mild cognitive impairment (MCI) subjects, a group at high risk for AD, and age-matched controls. A set of novel statistical methods and toolboxes will be developed to improve both dMRI image reconstruction and connectome estimation and analysis under acceleration. The collected data will be made publicly available. The project has three specific aims: (1) Determine the impact of accelerated imaging on structural connectome (SC) analysis in MCI subjects and age-matched controls. Subjects will have repeated scans under identical acquisition protocols. We will conduct quantitative assessments of reproducibility and discriminative ability of white matter structure and SC. (2) Machine learning reconstruction methods for reducing noise in accelerated dMRI. We will develop machine learning methods to improve vendor-implemented approaches to image reconstruction of k-space data to reduce noise and allow faster acquisition. (3) Develop novel methods for SC estimation and analysis in older adults. Utilizing the output from Aim 2, we will develop novel approaches that can better estimate and analyze SC for older adults. The methodologies developed in this project will facilitate the development of fast and high-quality dMRI acquisition and SC analysis and thus facilitate the development of early imaging biomarkers for AD.

Public Health Relevance

We aim to understand how accelerated imaging impacts dMRI data analysis in subjects at-risk of Alzheimer's disease (AD). We develop new machine learning methods for image reconstruction to overcome the drawbacks of accelerated acquisition. We develop robust structural connectome analysis methods for improved reproducibility. The data and algorithms generated in this study could lead to new practices in clinical neuroimaging of Alzheimer's disease (AD), improve our understanding of AD, and allow the development of new early biomarkers.

Agency
National Institute of Health (NIH)
Institute
National Institute on Aging (NIA)
Type
Exploratory/Developmental Grants (R21)
Project #
1R21AG066970-01
Application #
9952875
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Hsiao, John
Project Start
2020-05-01
Project End
2022-03-31
Budget Start
2020-05-01
Budget End
2021-03-31
Support Year
1
Fiscal Year
2020
Total Cost
Indirect Cost
Name
University of Rochester
Department
Biostatistics & Other Math Sci
Type
School of Medicine & Dentistry
DUNS #
041294109
City
Rochester
State
NY
Country
United States
Zip Code
14627