Neurodegenerative and Cerebrovascular Brain Ages Differentially Predict Cognition and Pathology
Poster No:
908
Submission Type:
Abstract Submission
Authors:
Yichi Zhang1, Fang Ji1, Joanna Su Xian Chong1, Joyce Chong1,2, Eric Kwun Kei Ng1, Nathanael Ren Jie Tong1, Susan Cheng1, Boon Yeow Tan3, Narayanaswamy Venketasubramanian2,4, Arthur Richards5,6, Mitchell Kim Peng Lai1,2, Michael Chee1, Thomas Yeo1,7,8,9, Christopher Chen1,2, Juan Helen Zhou1,7
Institutions:
1Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore, 2National University Health System, Singapore, Singapore, 3St Luke's Hospital, Singapore, Singapore, 4Raffles Neuroscience Centre, Raffles Hospital, Singapore, Singapore, 5Cardiovascular Research Institute, National University of Singapore, Singapore, Singapore, 6Christchurch Heart Institute, University of Otago, Dunedin, New Zealand, 7NUS Graduate School, National University of Singapore, Singapore, Singapore, 8N.1 Institute for Health, National University of Singapore, Singapore, Singapore, 9Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA
First Author:
Co-Author(s):
Joanna Su Xian Chong
Yong Loo Lin School of Medicine, National University of Singapore
Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore
Singapore, Singapore
Joyce Chong
Yong Loo Lin School of Medicine, National University of Singapore|National University Health System
Singapore, Singapore|Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore|National University Health System
Singapore, Singapore|Singapore, Singapore
Eric Kwun Kei Ng
Yong Loo Lin School of Medicine, National University of Singapore
Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore
Singapore, Singapore
Nathanael Ren Jie Tong
Yong Loo Lin School of Medicine, National University of Singapore
Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore
Singapore, Singapore
Narayanaswamy Venketasubramanian
National University Health System|Raffles Neuroscience Centre, Raffles Hospital
Singapore, Singapore|Singapore, Singapore
National University Health System|Raffles Neuroscience Centre, Raffles Hospital
Singapore, Singapore|Singapore, Singapore
Arthur Richards
Cardiovascular Research Institute, National University of Singapore|Christchurch Heart Institute, University of Otago
Singapore, Singapore|Dunedin, New Zealand
Cardiovascular Research Institute, National University of Singapore|Christchurch Heart Institute, University of Otago
Singapore, Singapore|Dunedin, New Zealand
Mitchell Kim Peng Lai
Yong Loo Lin School of Medicine, National University of Singapore|National University Health System
Singapore, Singapore|Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore|National University Health System
Singapore, Singapore|Singapore, Singapore
Thomas Yeo
Yong Loo Lin School of Medicine, National University of Singapore|NUS Graduate School, National University of Singapore|N.1 Institute for Health, National University of Singapore|Martinos Center for Biomedical Imaging, Massachusetts General Hospital
Singapore, Singapore|Singapore, Singapore|Singapore, Singapore|Charlestown, MA
Yong Loo Lin School of Medicine, National University of Singapore|NUS Graduate School, National University of Singapore|N.1 Institute for Health, National University of Singapore|Martinos Center for Biomedical Imaging, Massachusetts General Hospital
Singapore, Singapore|Singapore, Singapore|Singapore, Singapore|Charlestown, MA
Christopher Chen
Yong Loo Lin School of Medicine, National University of Singapore|National University Health System
Singapore, Singapore|Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore|National University Health System
Singapore, Singapore|Singapore, Singapore
Juan Helen Zhou, Ph.D.
Yong Loo Lin School of Medicine, National University of Singapore|NUS Graduate School, National University of Singapore
Singapore, Singapore|Singapore, Singapore
Yong Loo Lin School of Medicine, National University of Singapore|NUS Graduate School, National University of Singapore
Singapore, Singapore|Singapore, Singapore
Introduction:
Brain age has emerged as a powerful tool to understand neuroanatomical aging. Brain age gap (BAG) derived from T1 MRI summarizes complex patterns into a single number that preserves individual variations, which has been associated with cognition (Shah, 2003) and elevated AD-related biomarkers (Millar, 2023). Recent deep learning-based brain age models directly learn relevant features from minimally processed brain scans. However, there remain important gaps to address. First, cerebrovascular disorders (CeVD) are common in the aging brain and a major cause of cognitive decline and dementia (Wardlaw, 2013). Yet there is limited study on CeVD-related brain age models via FLAIR MRI. Second, most brain age models are trained on Caucasian cohorts. It is unclear how well they generalize to Asian cohorts. Therefore, here, we aimed to address these gaps by deriving both neurodegenerative (T1) and CeVD-related (FLAIR) BAGs from a longitudinal Singaporean memory cohort based on deep learning. We hypothesized that neurodegenerative BAG and CeVD-related BAG would differentially relate to blood biomarkers and predict future domain-specific cognitive decline. Specifically, we predicted that T1 BAG would be more related to neurodegeneration-based blood biomarker and memory decline, while FLAIR BAG would mainly relate to CeVD-related blood biomarkers and CeVD-impaired cognitive domains including executive function and processing speed.
Methods:
As shown in Figure 1, we finetuned the existing simple fully convolutional network (SFCN) (Leonardsen, 2022) brain age model on a healthy Singaporean elderly cohort SLABS (Chee, 2009) using either minimally processed T1 or FLAIR MRI based on 10-fold cross validations. We adopted Adam optimizer and finetuned both T1 and FLAIR brain age models for 35 epochs.
After finetuning, both brain age models were used to predict T1 and FLAIR BAGs on another Singaporean elderly dataset from memory clinic MACC-HARM (Xu, 2015), consisted of a spectrum of patients with cognitive impairment due to AD and/or CeVD. Apart from longitudinal MRI data, all the patients had baseline and longitudinal cognition scores of 7 domains and baseline plasma biomarkers AD-related PTau181, neurodegeneration-related NfL and CeVD-related hsTNT.
Correlation analyses were performed between both BAGs and all 7 domains of cognition scores (baseline and future rate of change) and baseline plasma biomarkers, controlling for chronological age, sex, education, race and cognitive status. Results were reported after multiple comparison correction.
After finetuning, both brain age models were used to predict T1 and FLAIR BAGs on another Singaporean elderly dataset from memory clinic MACC-HARM (Xu, 2015), consisted of a spectrum of patients with cognitive impairment due to AD and/or CeVD. Apart from longitudinal MRI data, all the patients had baseline and longitudinal cognition scores of 7 domains and baseline plasma biomarkers AD-related PTau181, neurodegeneration-related NfL and CeVD-related hsTNT.
Correlation analyses were performed between both BAGs and all 7 domains of cognition scores (baseline and future rate of change) and baseline plasma biomarkers, controlling for chronological age, sex, education, race and cognitive status. Results were reported after multiple comparison correction.
·Study design schematic
Results:
For MACC-HARM, the T1 finetuned brain age model achieved an MAE of 4.26 years, indicating good generalizability to Asian populations. Moreover, the MAE of the FLAIR finetuned model was 5.90 years, indicating that brain age models pretrained on T1 could also be finetuned to other modalities with similar brain anatomy.
Testing in separate models, both T1 and FLAIR BAGs were correlated with all 7 cognitive domains at baseline, future rates of change in cognition, and plasma biomarkers (PTau181, hsTNT and NfL) (p<0.05 corrected).
Importantly, when adding both BAGs in the same model, we found T1 BAG was more correlated with verbal and visual memory (Figure 2(A)), visual memory rate of change (Figure 2(E)) and AD-related PTau181 (Figure 2(C)). In contrast, FLAIR BAG was more correlated with visuomotor (Figure 2(B)), executive function rate of change (Figure 2(F)) and CeVD-related hsTNT level (Figure 2(D)).
Testing in separate models, both T1 and FLAIR BAGs were correlated with all 7 cognitive domains at baseline, future rates of change in cognition, and plasma biomarkers (PTau181, hsTNT and NfL) (p<0.05 corrected).
Importantly, when adding both BAGs in the same model, we found T1 BAG was more correlated with verbal and visual memory (Figure 2(A)), visual memory rate of change (Figure 2(E)) and AD-related PTau181 (Figure 2(C)). In contrast, FLAIR BAG was more correlated with visuomotor (Figure 2(B)), executive function rate of change (Figure 2(F)) and CeVD-related hsTNT level (Figure 2(D)).
·Brain age gap from the T1 and FLAIR finetuned models differentially relate to cognition scores and blood biomarkers
Conclusions:
Deep learning-based neurodegeneration (T1) and CeVD-related (FLAIR) BAGs differentially predict domain-specific cognition scores and pathology-specific biomarkers. Specifically, T1 BAG focused on memory decline and AD-related biomarkers while FLAIR BAG mainly related to CeVD impaired cognitive domains and cardiovascular biomarkers. Our findings highlight the importance of multimodal deep learning-based brain age models capturing both AD and CeVD pathology and different dimensions of cognitive decline.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Higher Cognitive Functions:
Executive Function, Cognitive Control and Decision Making
Lifespan Development:
Aging 1
Modeling and Analysis Methods:
Classification and Predictive Modeling
Keywords:
Aging
Blood
Cerebrovascular Disease
Cognition
Degenerative Disease
DISORDERS
Machine Learning
Memory
STRUCTURAL MRI
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.
Leonardsen, E. H. (2022). Deep neural networks learn general and clinically relevant representations of the ageing brain. NeuroImage, 256, 119210.
Millar, P. R. (2023). Advanced structural brain aging in preclinical autosomal dominant Alzheimer disease. Molecular neurodegeneration, 18(1), 98.
Shah, J. (2023). MRI signatures of Brain Age in the Alzheimer’s disease continuum. Alzheimer's & Dementia, 19, e063418.
Wardlaw, J. M. (2013). Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. The Lancet Neurology, 12(8), 822-838.
Xu, X. (2015). Association of magnetic resonance imaging markers of cerebrovascular disease burden and cognition. Stroke, 46(10), 2808-2814.