Using Personalized Functional Networks to Unravel the Signatures of Major Depressive Disorder
Poster No:
542
Submission Type:
Abstract Submission
Authors:
Alisha Kodibagkar1, Mathilde Antoniades1, Dhivya Srinivasan1, Cynthia Fu2, Christos Davatzikos3
Institutions:
1University of Pennsylvania, Philadelphia, PA, 2King's College London, London, N/a, 3Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA
First Author:
Co-Author(s):
Introduction:
Major Depressive Disorder (MDD) is increasingly understood as a network-level dysfunction[1]. Traditional group-averaged functional networks (FNs) from resting-state fMRI (rsfMRI) often miss individual variability, limiting insights into MDD's neural mechanisms. Personalized FNs (pFNs) account for individual brains' unique functional architecture[2-3]. Using Spatially Regularized Non-negative Matrix Factorization (SR-NMF), pFNs precisely identify individualized network signatures. This study investigates whether pFNs can improve MDD clinical phenotyping by distinguishing patients from controls and predicting response after Selective Serotonin Reuptake Inhibitor (SSRI) treatment.
Methods:
We analyzed baseline pretreatment rsfMRI scans from 913 subjects from the Coordinate-MDD consortium[4], including medication-naïve 1st-episode and medication-free recurrent MDD patients and controls. rsfMRI were preprocessed using fMRIPrep 23.1.3[5], which included T1-weighted image normalization and confound estimation, with BOLD time series resampled to native and MNI spaces. Post-processing with XCP-D[6] involved motion correction, despiking, detrending, and regression of 36 confounds, including motion and tissue signals. pFNs for all subjects were derived using the pNet[7] toolbox based on 17 group-level FNs from representative subjects in the Baltimore Longitudinal Study of Aging. pNet uses SR-NMF to identify distinct brain networks by decomposing BOLD signals into spatially smooth, functionally coherent components, enforcing shared structure across subjects, and generating pFNs with voxel-wise probabilities for network membership. Analyses of derived pFNs involved two comparisons: 1) MDD vs controls (497 MDD, 416 controls) and 2) MDD patients who respond to SSRI treatment vs those who do not (260 MDD responders vs 76 non-responders). For each comparison, voxel-wise loadings per network were summed to obtain total cortical representation[8], followed by two-sample t-tests with Bonferroni correction (p < 0.05) to assess group differences. Additionally, a logistic regression model with L2 regularization was trained on voxel-wise pFN loadings using 5-fold cross-validation to classify diagnostic or treatment groups, with network contributions calculated based on aggregated coefficients.
Results:
MDD subjects had median age 30(18–61), were 67% female, with median HAMD[9] and MADRS[10] scores of 20 and 27.5, onset age 19, and 8-month episode duration. Controls had median age 29(18–64), and were 60% female. In the MDD vs controls comparison, networks 10, 12, and 13 showed significant differences (corrected p=0.2, p=0.001, p<0.0001). The logistic regression model achieved a mean accuracy of 61.2%, log-loss of 1.55, and ROC-AUC of 0.66. Networks 12 and 13 negatively contributed to classification (associated with controls), while networks 14 and 15 positively contributed to MDD classification[Fig 1a]. In the treatment response analysis, network 15 showed significant differences between responders and non-responders (corrected p=0.05). The logistic regression model for treatment response classification achieved mean accuracy of 77.1%, log-loss of 1.13, and ROC-AUC of 0.71. Networks 15 and 7 positively contributed to treatment response prediction, while networks 9 and 10 negatively contributed[Fig 1b]. Example networks for a control and MDD patient[Fig 2a-b], and for responder and non-responder patients[Fig 2c-d], highlight significant networks.
Conclusions:
This study shows pFNs effectively identify individualized brain network alterations in MDD. For MDD vs control, significant differences in networks 10, 12, and 13, and the predictive contributions of networks 14 and 15, highlight distinct neural signatures of MDD. For clinical treatment response, significant differences in network 15 and the predictive contributions of networks 7, 9, and 10 highlight distinct neural signatures of MDD response to treatment. The findings advance understanding of MDD as a network dysfunction.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Modeling and Analysis Methods:
fMRI Connectivity and Network Modeling 2
Task-Independent and Resting-State Analysis
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
FUNCTIONAL MRI
Psychiatric
Psychiatric Disorders
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?
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
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:
Provide references using APA citation style.
[2] Li, H., Satterthwaite, T. D., & Fan, Y. (2017). Large-scale sparse functional networks from resting state fMRI. Neuroimage, 156, 1-13.
[3] Cui, Z., Li, H., Xia, C. H., Larsen, B., Adebimpe, A., Baum, G. L., ... & Satterthwaite, T. D. (2020). Individual variation in functional topography of association networks in youth. Neuron, 106(2), 340-353.
[4] Fu, C. H., Erus, G., Fan, Y., Antoniades, M., Arnone, D., Arnott, S. R., ... & Davatzikos, C. (2023). AI-based dimensional neuroimaging system for characterizing heterogeneity in brain structure and function in major depressive disorder: COORDINATE-MDD consortium design and rationale. BMC psychiatry, 23(1), 59.
[5] Esteban, O., Markiewicz, C. J., Blair, R. W., Moodie, C. A., Isik, A. I., Erramuzpe, A., ... & Gorgolewski, K. J. (2019). fMRIPrep: a robust preprocessing pipeline for functional MRI. Nature methods, 16(1), 111-116.
[6] Mehta, K., Salo, T., Madison, T. J., Adebimpe, A., Bassett, D. S., Bertolero, M., ... & Satterthwaite, T. D. (2024). XCP-D: A Robust Pipeline for the post-processing of fMRI data. Imaging Neuroscience, 2, 1-26.
[7] Ma, Y., Li, H., Zhou, Z., Chen, X., Ma, L., Guray, E., ... & Fan, Y. (2024). pNet: A toolbox for personalized functional networks modeling. bioRxiv.
[8] Keller, A. S., Pines, A. R., Shanmugan, S., Sydnor, V. J., Cui, Z., Bertolero, M. A., ... & Satterthwaite, T. D. (2023). Personalized functional brain network topography is associated with individual differences in youth cognition. Nature communications, 14(1), 8411.
[9] Hamilton, M. (1960). A rating scale for depression. Journal of neurology, neurosurgery, and psychiatry, 23(1), 56.
[10] Montgomery, S. A., & Åsberg, M. A. R. I. E. (1979). A new depression scale designed to be sensitive to change. The British journal of psychiatry, 134(4), 382-389.