Poster No:
222
Submission Type:
Abstract Submission
Authors:
Konstantinos Poulakis1,2, Rosaleena Mohanty1, Anna Inguanzo1, Eric Westman1,3
Institutions:
1Karolinska Institutet, Stockholm, Sweden, 2McGill University, Montreal, Quebec, Canada, 3Kings College London, London, United Kingdom
First Author:
Co-Author(s):
Eric Westman
Karolinska Institutet|Kings College London
Stockholm, Sweden|London, United Kingdom
Introduction:
Human brain grey matter atlases, based on sulci and/or gyri hallmarks, are used to parcellate the human brain into regions of interest (ROIs) and thus reduce the information and noise inherent in brain imaging to a few relevant features. Although ROI measures are employed in research to understand neurological diseases, it is unknown whether these parcellations consider potential disease-specific atrophy patterns. Moreover, a single central tendency ROI statistic (mean or median) is often used, ignoring the potentially important within ROI dispersion, e.g. standard deviation (SD) or median absolute deviation.
Methods:
Utilizing a multi-view latent factor model for multiple dataset exploration (joint and individual variation explained (Lock et al., 2013)), we investigated the relationship between the mean and SD of 148 grey matter thickness ROI measures (right/left hemisphere) in the context of cognitively unimpaired healthy aging (CU, n = 471), mild cognitive impairment (MCI, n= 339), Alzheimer's disease (AD, n = 336), and Parkinson's disease (PD, n = 324). We utilized cross-sectional T1 magnetic resonance imaging data from the ADNI (CU, MCI, AD), J-ADNI (CU, AD), AIBL (CU, AD), and PPMI (CU, PD) cohorts. Furthermore, longitudinal global and specific neuropsychological measures of cognitive and motor health (Alzheimer's disease assessment scale-cog, mini mental state examination, clinical dementia rating scale, Montreal cognitive assessment, neuro QoL: cognition function - short form, modified Boston naming test total correct, clock drawing total score, rapid eye movement symptoms, Hoehn and Yahr Stage) were correlated with the discovered ROI-based latent factors.
Results:
Mean and SD grey matter thickness at the ROI level exhibited common but also independent patterns of variation, showcasing that each statistic retrieves different disease-related anatomical information. The CU, MCI, and AD datasets exhibited similarities between their grey matter latent factor estimates. The CU, MCI, and AD latent factors increasingly correlated with cognition with the advancement of clinical progression . Interestingly, although the CU, MCI, and AD groups presented both common and individual variation in their mean and SD signals, the PD group showed only individual mean and SD variation signals. Moreover, the patterns of atrophy (latent atrophy factors based on mean and SD cortical thickness) in the MCI, AD, and PD groups were different .
Conclusions:
The SD of grey matter thickness shares common features with mean grey matter thickness but also provides unique disease-related information in the healthy and cognitively impaired populations. SD is complementary to mean grey matter thickness in predicting future cognitive decline. AD and PD differ significantly in their mean-SD grey matter thickness dynamics, showcasing the potential of combined atrophy descriptive markers for differentiating and exploring these diseases.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1
Lifespan Development:
Aging
Modeling and Analysis Methods:
Bayesian Modeling
Multivariate Approaches 2
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping
Keywords:
Aging
Multivariate
Statistical Methods
Other - Multiview data analysis
1|2Indicates the priority used for review
Provide references using author date format
E. F. Lock, K. A. Hoadley, J. S. Marron, and A. B. Nobel. (2013), “Joint and individual variation explained (JIVE) for integrated analysis of multiple data types,” Ann. Appl. Stat., vol. 7, no. 1, doi: 10.1214/12-AOAS597.