Functional Fingerprinting for the Developing Brain Using Deep Metric Learning: An ABCD Study
Presented During:
Thursday, June 26, 2025: 11:30 AM - 12:45 PM
Brisbane Convention & Exhibition Centre
Room:
Great Hall
Poster No:
1014
Submission Type:
Abstract Submission
Authors:
Rui Xu1, Yongbin Wei1, Ting Qi1, Shuwan Zhao1, Yong Liu1
Institutions:
1School of Artificial Intelligence,Beijing University of Posts and Telecommunications, Beijing, China
First Author:
Rui Xu
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
Co-Author(s):
Yongbin Wei
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
Ting Qi
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
Shuwan Zhao
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
Yong Liu
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence,Beijing University of Posts and Telecommunications
Beijing, China
Introduction:
Individual differences in brain function have been prominently observed during development. To characterize this uniqueness, researchers use functional magnetic resonance imaging (fMRI) based brain fingerprinting, achieving high-precision individual identification in adults (Finn, 2015). Current methods demonstrate great potential in capturing functional fingerprints, yet more refined methodologies are needed to delineate the unique developmental trajectory of the brain. Here, we propose a deep learning framework for longitudinal functional fingerprinting in early adolescents, identifying unique brain functional patterns relating to cognitive abilities and genetics.
Methods:
Minimally processed T1-weighted MRI and resting-state fMRI data from the Adolescent Brain Cognitive Development (ABCD) study were used (data release 5.1; DAR ID: 16920) (Garavan, 2018), with scans at baseline, second-year and fourth-year follow-up. The T1-weighted data were processed with FreeSurfer (v.6.0) for brain segmentation, parcellating the cortical ribbon into 219 regions using the DK-219 atlas (Cammoun, 2012; Desikan, 2006). Resting-state fMRI data were processed with CATO (v.3.1.2) (de Lange, 2023): realignment, co-registration with T1, motion regression, and bandpass filtering (0.01 to 0.1 Hz).
We propose Metric-BolT (Bedel, 2023), a distance-based metric learning approach that encodes time series into brain fingerprints while maximizing inter-subject and minimizing intra-subject fingerprint distances. Individual identification performed at all time points and intervals; statistical significance determined by comparing observed values to null distributions from non-parametric permutation tests (1,000 iterations). Contributions of brain regions are quantified by analyzing model coefficients W and are annotated using Neuromaps (Markello, 2022). Associations between brain fingerprints and behavioral abilities are examined using least squares regression and F-tests. We compared the distances between brain fingerprints of subjects with strong genomic relationships to those between unrelated subjects.
We propose Metric-BolT (Bedel, 2023), a distance-based metric learning approach that encodes time series into brain fingerprints while maximizing inter-subject and minimizing intra-subject fingerprint distances. Individual identification performed at all time points and intervals; statistical significance determined by comparing observed values to null distributions from non-parametric permutation tests (1,000 iterations). Contributions of brain regions are quantified by analyzing model coefficients W and are annotated using Neuromaps (Markello, 2022). Associations between brain fingerprints and behavioral abilities are examined using least squares regression and F-tests. We compared the distances between brain fingerprints of subjects with strong genomic relationships to those between unrelated subjects.
Results:
The Success Rate (SR) reached 93.4% at baseline, 95.2% and 97.3% at second-year and fourth-year follow-up respectively (Fig. 1A). Longitudinal identification showed SR of 90.9% for 2-year and 86.6% for 4-year interval, outperforming FC-based methods. The ratio of average intra-class to inter-class distances (CR) increased from 0.41(within-session) to 0.47 (2-year) and 0.5 (4-year), while the silhouette coefficients (SC) decreased from 0.45 to 0.34 and 0.30 (Fig. 1B). Permutations tests yielded max SR of 4.07%, 5.00%, and 4.59% (all p<0.001) (Fig. 1C). Genomically related subjects showed smaller brain fingerprint distances across three intervals (t=-12.3, -7.5, -9.1; p<0.001). High correlations in regional contributions to brain fingerprints were observed between within-session, 2-year, and 4-year interval identifications (all r > 0.94, p < 0.001; Fig. 2A,B). Common regions include bilateral middle temporal, superior frontal, and inferior parietal regions. Among functional networks, regions from the DMN showed the highest levels of contributions, significantly higher than other functional networks (t(217) = 5.959, p < 0.001) and higher-order cognitive networks (DMN, FPN, SN) showed significantly higher levels than primary networks (VN,SMN) (t(176) = 5.618, p < 0.001; Fig. 2C). Regional contribution patterns also correlates to the fcgradient (Fig. 2D). Brain fingerprints showed significant associations with fluid intelligence (F = 1.282, p = 0.027) and crystallized intelligence (F = 1.405, p < 0.001).
·Individual identification results across various time intervals
·Regional contributions to the extracted brain fingerprints.
Conclusions:
We developed a deep metric learning framework for longitudinal brain fingerprinting in early adolescents. This framework showed a high precision in individual identification over four years and highlighted the individual variability in higher-order association networks. Our approach offers new insights into unique brain developmental trajectories in early adolescence.
Lifespan Development:
Normal Brain Development: Fetus to Adolescence 1
Lifespan Development Other
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI)
Classification and Predictive Modeling 2
Keywords:
FUNCTIONAL MRI
Other - brain fingerprinting; individual variability; brain development; fMRI; individual identification
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.
Cammoun, L., et al. (2012). Mapping the human connectome at multiple scales with diffusion spectrum MRI. Journal of Neuroscience Methods, 203(2), 386-397.
de Lange, S. C., et al. (2023). Structural and functional connectivity reconstruction with CATO - A Connectivity Analysis TOolbox. NeuroImage, 273, 120108.
Desikan, R. S., et al. (2006). An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage, 31(3), 968-980.
Finn, E. S., et al. (2015). Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity. Nature Neuroscience, 18(11), 1664-1671.
Garavan, H., et al. (2018). Recruiting the ABCD sample: Design considerations and procedures. Developmental Cognitive Neuroscience, 32, 16-22.
Markello, R. D., et al. (2022). neuromaps: structural and functional interpretation of brain maps. Nature Methods, 19(11), 1472-1479.