Spectral Normative Modeling (SNM) for High-Resolution Brain Abnormality Inference
Presented During:
Wednesday, June 25, 2025: 5:45 PM - 7:00 PM
Brisbane Convention & Exhibition Centre
Room:
M3 (Mezzanine Level)
Poster No:
200
Submission Type:
Abstract Submission
Authors:
Sina Mansour L.1, Maria Di Biase2, Yan Hongwei3, Aihuiping Xue4, Narayanaswamy Venketasubramanian5, Eddie Chong6, Christopher Chen3, Helen Juan Zhou3, Thomas Yeo7, Andrew Zalesky8
Institutions:
1University of Melbourne & National University of Singapore, Melbourne, Australia, 2The University of Melbourne, Melbourne, VIC, 3National University of Singapore, Singapore, Singapore, 4Computational Brain Imaging Group, Yong Loo Lin School of Medicine, National University of Singapor, Singapore, Singapore, 5National University Health System, Singapore, Singapore, 6National University Health System, National University of Singapore, Singapore, Singapore, 7Centre for Sleep and Cognition, National University of Singapore, Singapore, Singapore, 8Systems Lab, Department of Psychiatry, The University of Melbourne, Melbourne, Australia
First Author:
Sina Mansour L., Ph.D.
University of Melbourne & National University of Singapore
Melbourne, Australia
University of Melbourne & National University of Singapore
Melbourne, Australia
Co-Author(s):
Aihuiping Xue
Computational Brain Imaging Group, Yong Loo Lin School of Medicine, National University of Singapor
Singapore, Singapore
Computational Brain Imaging Group, Yong Loo Lin School of Medicine, National University of Singapor
Singapore, Singapore
Eddie Chong
National University Health System, National University of Singapore
Singapore, Singapore
National University Health System, National University of Singapore
Singapore, Singapore
Andrew Zalesky
Systems Lab, Department of Psychiatry, The University of Melbourne
Melbourne, Australia
Systems Lab, Department of Psychiatry, The University of Melbourne
Melbourne, Australia
Introduction:
Normative models (NMs) in neuroscience aim to characterize interindividual variability in brain phenotypes, establishing reference ranges-or brain charts-against which individual brains can be compared. NMs help identify individual-level abnormalities as potential biomarkers for neurological and psychiatric disorders [1,2,3]. However, conventional NMs face computational challenges in mapping charts at high spatial resolutions. State-of-the-art techniques [4,5] rely on exhaustive model fitting for each voxel or vertex, which is computationally burdensome and limits scalability for large neuroimaging datasets.
The high dimensionality of neuroimaging data complicates the development of high-resolution NMs. Identifying low-dimensional representations of cortical features can help overcome this challenge. Recent works suggest that brain eigenmodes may provide effective low-dimensional basis functions for reconstructing phenotypic variation on the cortical surface [6,7]. Here, we introduce spectral normative modeling (SNM), a novel approach that leverages spatial reconstructions via brain eigenmodes to generate normative brain charts at varying spatial scales and resolutions. By computing normative ranges of eigenmodes and accounting for cross-mode dependencies, SNM offers an efficient solution for high-resolution charting of brain phenotypes.
Alzheimer's disease (AD) is a neurodegenerative disorder marked by the accumulation of amyloid-beta plaques and tau tangles, contributing to neuronal loss and cortical atrophy, leading to progressive structural changes in brain regions critical for memory and cognition, differentiating AD brains from the normative trajectories of healthy aging [8]. We applied SNM to identify these structural deviations at the individual level, demonstrating its utility in characterizing neurological disorders and identifying potential prognostic biomarkers.
The high dimensionality of neuroimaging data complicates the development of high-resolution NMs. Identifying low-dimensional representations of cortical features can help overcome this challenge. Recent works suggest that brain eigenmodes may provide effective low-dimensional basis functions for reconstructing phenotypic variation on the cortical surface [6,7]. Here, we introduce spectral normative modeling (SNM), a novel approach that leverages spatial reconstructions via brain eigenmodes to generate normative brain charts at varying spatial scales and resolutions. By computing normative ranges of eigenmodes and accounting for cross-mode dependencies, SNM offers an efficient solution for high-resolution charting of brain phenotypes.
Alzheimer's disease (AD) is a neurodegenerative disorder marked by the accumulation of amyloid-beta plaques and tau tangles, contributing to neuronal loss and cortical atrophy, leading to progressive structural changes in brain regions critical for memory and cognition, differentiating AD brains from the normative trajectories of healthy aging [8]. We applied SNM to identify these structural deviations at the individual level, demonstrating its utility in characterizing neurological disorders and identifying potential prognostic biomarkers.
Methods:
To demonstrate SNM's utility, we trained it on a healthy cohort to model lifelong normative cortical thickness development as a function of age and sex covariates. Data from the Human Connectome Project lifespan datasets (N = 2,473) were used for model training [9,10,11]. We benchmarked SNM against conventional approaches in terms of accuracy and computational efficiency. Furthermore, we leverage SNM's transfer learning capability to fine-tune the model on an independent clinical dataset comprising three elderly cohorts: healthy controls (HC, N = 132), individuals with mild cognitive impairments (MCI, N = 202), and those with Alzheimer's disease (AD, N = 208) [12]. The resulting high-resolution normative deviation maps (z-scores) were used to study cortical abnormalities associated with AD dementia.
Results:
SNM achieved a 98.3% reduction in computational time compared to conventional methods while maintaining comparable accuracy (Fig. 1). Utilizing SNM's ability to efficiently map high-resolution normative references, we investigated the cortical atrophy signatures associated with AD dementia uncovering widespread atrophy spanning the temporal, parietal, and frontal areas (Fig. 2). Moreover, we identified a normative biomarker of extreme cortical atrophy predictive of AD-related cognitive impairments. By examining intersubject variability in normative deviations, we uncovered a cortical atrophy landscape, situating AD dementia as extreme and heterogeneous deviations from a majorly healthy centroid, shedding light on the individually diverse nature of neurodegeneration in AD dementia.
Conclusions:
Our findings underscore SNM's potential to advance individualized precision medicine by enabling the next generation of fine-grained brain charts. SNM offers nuanced normative evaluations that surpass traditional group-level approaches, providing a powerful tool to study cortical abnormalities at the individual level. Although AD serves as an exemplary use case, the utility of SNM is extensible to other neurological and psychiatric disorders.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1
Lifespan Development:
Aging
Early life, Adolescence, Aging
Modeling and Analysis Methods:
Bayesian Modeling
Methods Development 2
Keywords:
Aging
Computational Neuroscience
Cortex
Degenerative Disease
Design and Analysis
Machine Learning
NORMAL HUMAN
Open-Source Code
STRUCTURAL MRI
Other - Normative modeling
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
[2] Rutherford, Saige, et al. "The normative modeling framework for computational psychiatry." Nature protocols 17.7 (2022): 1711-1734.
[3] Bethlehem, Richard AI, et al. "Brain charts for the human lifespan." Nature 604.7906 (2022): 525-533.
[4] Holz, Nathalie E., et al. "A stable and replicable neural signature of lifespan adversity in the adult brain." Nature Neuroscience 26.9 (2023): 1603-1612.
[5] Wolfers, Thomas, et al. "Individual differences v. the average patient: mapping the heterogeneity in ADHD using normative models." Psychological medicine 50.2 (2020): 314-323.
[6] Pang, James C., et al. "Geometric constraints on human brain function." Nature 618.7965 (2023): 566-574.
[7] Mansour L, Sina, et al. "Eigenmodes of the brain: revisiting connectomics and geometry." bioRxiv (2024): 2024-04.
[8] Hampel, Harald, et al. "Developing the ATX (N) classification for use across the Alzheimer disease continuum." Nature Reviews Neurology 17.9 (2021): 580-589.
[9] Somerville, Leah H., et al. "The Lifespan Human Connectome Project in Development: A large-scale study of brain connectivity development in 5–21 year olds." Neuroimage 183 (2018): 456-468.
[10] Van Essen, David C., et al. "The WU-Minn human connectome project: an overview." Neuroimage 80 (2013): 62-79.
[11] Bookheimer, Susan Y., et al. "The lifespan human connectome project in aging: an overview." Neuroimage 185 (2019): 335-348.
[12] Hilal, Saima, et al. "Cortical cerebral microinfarcts predict cognitive decline in memory clinic patients." Journal of Cerebral Blood Flow & Metabolism 40.1 (2020): 44-53.