A normative model of cortical morphology across the lifespan
Poster No:
2145
Submission Type:
Abstract Submission
Authors:
Bethany Little1, Karoline Leiberg2, Nida Alyas1, Yujiang Wang3
Institutions:
1Newcastle University, Newcastle upon Tyne, United Kingdom, 2Newcastle University, Newcastle upon Tyne, Tyne and Wear, 3Newcastle University, Newcastle, United Kingdom
First Author:
Co-Author(s):
Introduction:
Understanding the intricate morphology of the human brain in health and disease is a foundational pursuit in neuroscience and current challenges in this field stem from noisy imaging data and small sample sizes, especially for clinical studies. Traditional measures of cortical morphology, i.e. thickness, volume, and surface area, are known to covary and do not capture the complex folded shape of the brain. Research is beginning to leverage normative modelling as a framework for creating robust estimates of healthy variations in brain structure across the lifespan and to assess abnormalities in patient cohorts or individuals. We aimed to extend this field by developing a normative model of brain morphology that can estimate variations in traditional structural metrics as well as novel measures of cortical morphology, and make this model available to the community.
Methods:
We collated T1-weighted MRI data from several large public datasets (including HCP, NKI, OASIS, and CamCAN) and in-house studies of healthy controls across the lifespan (n>3,500, age range 6-95 years). All data were pre-processed using the standard recon-all pipeline in FreeSurfer. We utilised novel independent components of cortical morphology (Wang et al., 2020, NeuroImage) to provide nuanced measures of brain structure. We used generalized additive models to build normative models of the independent components (K, I, and S) as well as traditional measures, accounting for normative age, sex, and site effects.
Results:
The normative models of traditional metrics showed age trends in line with previous research, e.g. decreases in cortical thickness with age. The independent morphological metric K also decreased with age and explained more of the variance in the data than the traditional metrics, suggesting K may be a more appropriate measure to describe ageing and detect deviations from the healthy trajectory. We introduce our analysis pipeline as a freely available web app that can take new data as input and estimate abnormalities in cortical morphology for each individual based on the normative data. Crucially, we demonstrate that, given a new dataset with a clinical group and matched healthy controls, our model can estimate abnormalities in each patient, where biological and technical covariates are corrected based on the healthy data, debiasing site effects and more accurately estimating the underlying psychopathology of clinical populations.
Conclusions:
Our normative models provide robust estimations of healthy variations, and abnormalities, in brain structure across the lifespan, utilising both traditional and novel metrics. Our freely available web app offers an accessible and powerful tool to estimate nuanced measures of cortical morphology and will open new avenues for research and clinical applications for detecting brain structural abnormalities.
Lifespan Development:
Lifespan Development Other
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping
Normal Development 1
Neuroinformatics and Data Sharing:
Databasing and Data Sharing 2
Informatics Other
Keywords:
Computational Neuroscience
Cortex
Data analysis
Design and Analysis
Modeling
Morphometrics
MRI
NORMAL HUMAN
STRUCTURAL MRI
Structures
1|2Indicates the priority used for review
·Normative models of cortical thickness (top panel; in log space) and K (bottom panel) across the lifespan for males and females.