Poster No:
1172
Submission Type:
Abstract Submission
Authors:
Yae Ji Kim1, Minchul Kim2, Marvin Chun3, Kwangsun Yoo4
Institutions:
1Sungkyunkwan University, Gangnam-gu, Seoul, 2Kangbuk Samsung Hospital, Seoul, Seoul, 3Department of Psychology, Yale University, , CT, USA, New Haven, CT, 4Sungkyunkwan University, ., Seoul
First Author:
Yae Ji Kim
Sungkyunkwan University
Gangnam-gu, Seoul
Co-Author(s):
Marvin Chun
Department of Psychology, Yale University, , CT, USA
New Haven, CT
Introduction:
In clinical research and practice, structural similarity mapping can estimate an individual's structural brain network from structural T1 MRI (Sebenius et al., 2024). However, its utility as a proxy for functional connectomes and its predictive power for behavior have not been fully validated. To explore this potential, we adapted the connectome-to-connectome (C2C) transformation modeling framework to transform structural similarity matrices into whole brain functional connectomes (FCs), further improving their predictive power comparable to empirical FCs (Yoo et al., 2022b).
Methods:
We used an fMRI dataset from Yoo et al., 2022a, including resting-state and attention task (gradual onset continuous performance task [gradCPT]) fMRI data from 92 participants. Structural similarity networks (SSNs) were derived using z-transformed histogram correlate metrics between volumetric regions parcellated by the Shen atlas (Oskar et al., 2019; Shen et al., 2013). Using transformation modeling, we built a computational model that predicts task-free FCs from SSNs (Yoo et al., 2022b). Our model consists of three steps: (1) extracting structural subcomponents from the SSNs, (2) transforming them into task-free functional subcomponents, and (3) constructing whole-brain task-free FCs based on the transformed subcomponents. We evaluated the model's ability to predict task-free FCs by comparing the predicted FCs with empirical task-free FCs. High similarity between the FCs was considered a successful prediction, assessed using spatial correlation (r), mean squared error (MSE), and q² values. Brain state specificity was assessed using state fingerprinting. We computed identification success rates (%) to evaluate if predicted task-free FCs were most similar to the corresponding empirical task-free FCs, compared to the other task-related FCs. Finally, we examined whether the predicted task-free FCs from SSNs predict an individual's attention functions (gradCPT and general attention) using connectome-based predictive modeling (CPM) (Yoo et al., 2022a). We trained and validated the C2C and CPM using a 10-fold cross-validation (CV) with 1000 iterations (Yoo et al., 2022b).
Results:
We observed high spatial correlation between predicted task-free FCs and empirical task-free FCs (r = 0.70; MSE = 0.02; q2 = 0.49), demonstrating successful prediction of FCs from SSNs. The similarity between predicted and empirical task-free FCs was significantly higher than the similarity between empirical SSNs and empirical task-free FCs (p < 0.01). Our model also provided high state specificity, demonstrated by an identification success rate of 91.3%.
Behavioral predictions using predicted task-free FCs (gradCPT: r = 0.28, p < 0.01, q2 = 0.07; general attention: r = 0.22, p = 0.04, q2 = 0.05) significantly surpassed predictions using SSNs alone in gradCPT and general attention measures (gradCPT: r = 0.08, p = 0.43, q2 = -0.13; general attention: r = 0.06, p = 0.56, q2 = -0.13), and even exhibited comparable prediction performance to empirical task-free FCs (gradCPT: r = 0.39, p < 0.01, q2 = 0.13; general attention: r = 0.382 p < 0.01, q2 = 0.12).
Conclusions:
In this study, we introduced a cross-modal connectome transformation model to explore the potential of structural MRI-based single-subject brain networks to predict behavior. The model successfully predicted task-free FCs from SSNs, where the predicted task-free FCs outperformed the empirical structural networks in predicting an individual's attentional behaviors, even comparable to predictions from empirical task-free FCs. The results highlight that single-subject SSNs may encode functionally relevant information, suggesting that when integrated with the cross-modal connectome transformation model, they can provide a practical solution when direct functional data is unavailable.
Modeling and Analysis Methods:
Classification and Predictive Modeling
Connectivity (eg. functional, effective, structural) 1
Novel Imaging Acquisition Methods:
Anatomical MRI
Multi-Modal Imaging 2
Perception, Attention and Motor Behavior:
Perception and Attention Other
Keywords:
Computational Neuroscience
FUNCTIONAL MRI
STRUCTURAL MRI
Other - attention
1|2Indicates the priority used for review
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.
I accept
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:
I do not want to participate in the reproducibility challenge.
Please indicate below if your study was a "resting state" or "task-activation” study.
Resting state
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Healthy subjects
Was this research conducted in the United States?
No
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.
Yes
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.
Not applicable
Please indicate which methods were used in your research:
Functional MRI
Structural MRI
Behavior
Computational modeling
For human MRI, what field strength scanner do you use?
3.0T
Provide references using APA citation style.
Oskar Maier, Alex Rothberg, Pradeep Reddy Raamana, Rémi Bèges, Fabian Isensee, Michael Ahern, mamrehn, VincentXWD, & Jay Joshi. (2019). loli/medpy: MedPy 0.4.0 (0.4.0). Zenodo. https://doi.org/10.5281/zenodo.2565940
Sebenius, I., Dorfschmidt, L., Seidlitz, J., Alexander-Bloch, A., Morgan, S. E., & Bullmore, E. (2024). Structural MRI of brain similarity networks. Nature reviews. Neuroscience, 10.1038/s41583-024-00882-2. Advance online publication. https://doi.org/10.1038/s41583-024-00882-2
Shen, X., Tokoglu, F., Papademetris, X., & Constable, R. T. (2013). Groupwise whole-brain parcellation from resting-state fMRI data for network node identification. NeuroImage, 82, 403–415. https://doi.org/10.1016/j.neuroimage.2013.05.081
Yoo, K., Rosenberg, M. D., Kwon, Y. H., Lin, Q., Avery, E. W., Sheinost, D., Constable, R. T., & Chun, M. M. (2022a). A brain-based general measure of attention. Nature human behaviour, 6(6), 782–795. https://doi.org/10.1038/s41562-022-01301-1
Yoo, K., Rosenberg, M. D., Kwon, Y. H., Scheinost, D., Constable, R. T., & Chun, M. M. (2022b). A cognitive state transformation model for task-general and task-specific subsystems of the brain connectome. NeuroImage, 257, 119279. https://doi.org/10.1016/j.neuroimage.2022.119279
No