Abstract #4758
Diffusion MRI harmonization and thresholding improve multicentre network analysis: a demonstration in cerebral small vessel disease
Bruno Miguel de Brito Robalo1,2, Alberto de Luca1,2, Christopher Chen3, Anna Dewenter4, Marco Duering5, Saima Hilal3, Huiberdina L. Koek6, Anna Kopczak4, Bonnie Yin Ka Lam7, Alexander Leemans2, Vincent Mok7, Laurien P. Onkenhout1, Hilde van den Brink1, and Geert Jan Biessels1
1Neurology, Department of Neurology and Neurosurgery, UMC Utrecht Brain Center, University Medical Center Utrecht, Utrecht, Netherlands, 2Image Sciences Institute, University Medical Center Utrecht, Utrecht University, Utrecht, Netherlands, 3Memory, Aging and Cognition Center, Department of Pharmacology, National University of Singapore, Singapore, Singapore, 4Institute for Stroke and Dementia Research (ISD), University Hospital, LMU Munich, Munich, Germany, 5Medical Image Analysis Center (MIAC) and qbig, Department of Biomedical Engineering, University of Basel, Basel, Switzerland, 6Department of Geriatric Medicine, University Medical Center Utrecht, Utrecht, Netherlands, 7Department of Geriatric Medicine, University Medical Center Utrecht, Utrecht, The Netherlands 7Division of Neurology, Department of Medicine and Therapeutics, Gerald Choa Neuroscience Centre, Faculty of Medicine, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong, Hong Kong
Synopsis
We investigated if network thresholding and diffusion MRI (dMRI) harmonization improve a) cross-site consistency of network architecture and b) precision and sensitivity to detect network connections disrupted in cerebral small vessel disease (SVD). Brain networks were reconstructed from dMRI in five cohorts. Consistency of network architecture was examined in age-matched controls whereas sensitivity and precision to detect disrupted connections was assessed in sporadic SVD patients. Network consistency, as well as sensitivity and precision to detect disrupted connections were improved by thresholding and harmonization. We recommend using these techniques in networks studies of SVD to leverage existing multicentre datasets.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
