Diffeomorphic signal alignment for fMRI multivoxel pattern analysis
Poster No:
1067
Submission Type:
Abstract Submission
Authors:
Brandon Taraku1, David Lee2, Katherine Narr1, Shantanu Joshi1
Institutions:
1UCLA, Los Angeles, CA, 2UCLA Health, Los Angeles, CA
First Author:
Co-Author(s):
Introduction:
Functional MRI (fMRI) decoding techniques, such as multivoxel pattern analysis (MVPA), have revealed fine-grained functional representations in task-based fMRI studies, achieving greater sensitivity than traditional univariate approaches (Weaverdyck et al., 2020). However, MVPA based approaches can suffer from poor decoding accuracies when not taking into account functional heterogeneity across subjects. Approaches such as hyperalignment (Haxby et al., 2020) have been proposed to align functional patterns across individuals, but this approach can have some limitations (Andreella et al., 2023), and doesn't consider diffeomorphisms. Here, we apply a novel fMRI alignment approach described in (Lee et al., 2020), to align subjects for MVPA. This approach incorporates a global warping function that accounts for phase changes across all signals being aligned (temporal reparameterizations), and can be combined with temporal rotations (Joshi et al., 2018) to account for changes in signal magnitude. These alignment approaches and hyperalignment were thus used to determine how they affect the decoding of distinct fMRI task-based stimuli across subjects. A representational similarity analysis (RSA) (Weaverdyck et al., 2020) was then used to compare the similarity of task conditions between subjects with major depressive disorder (MDD) and healthy controls (HC) to determine how alignment impacts the detection of group differences.
Methods:
Participants included 53 HC with no diagnosis of psychiatric or neurological conditions (mean age = 32.74, female = 29) and 53 participants with MDD (mean age = 36.29, female = 28). All participants received fMRI scans while performing a validated emotion recognition face-matching task (Loureiro et al., 2020), where they viewed stimuli including happy, fearful, and neutral faces, and objects. Imaging data was preprocessed using the Human Connectome Project minimal preprocessing pipeline (Glasser et al., 2013). Regions of interest (ROIs) chosen for further analysis included the left and right amygdala (Amyg) and fusiform face area (FFA). Functional alignments were performed on each of the ROIs separately across all subjects. Hyperalignment was calculated using the NLTools library in Python (Chang et al., 2018), and the diffeomorphic alignment techniques were calculated using custom MATLAB code. MVPA was performed using a linear support vector machine (SVM) in Python to classify patterns of activity across each ROI, using the set of voxels as features and the stimulus type as class labels. RSA was performed in Python to compute the voxelwise similarity of task conditions in each ROI, using Pearson correlations. Differences in similarity across conditions between MDD and HC were determined using independent samples t-tests.
Results:
SVM test accuracies for each alignment and ROI are displayed in Figure 1, along with SVM weights visualized on the brain.
RSA results are displayed in Figure 2.
RSA results are displayed in Figure 2.
Conclusions:
Our SVM results show that functional alignment enhances neural representations of emotional stimulus classification during fMRI. While hyperalignment performs well, temporal rotations and reparameterizations outperform it across all tasks. This novel alignment approach may thus be optimal for fMRI decoding. When weights of the classifier are visualized, discernible patterns in the FFA and amygdala can be observed, suggesting that despite using different alignment approaches, similar underlying brain representations appear to be decoded. RSA results suggest that this novel alignment approach may be more sensitive for detecting differences across diagnostic groups within multivariate brain patterns. Specifically, our RSA results show greater dissimilarity between happy and fearful faces in the amygdala, suggesting that MDD have exaggerated brain responses to emotional stimuli. This is in line with prior research which found hyperactivity of amygdala responses for affective stimuli in MDD (Fournier et al., 2012; Stuhrmann et al., 2011).
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia)
Emotion, Motivation and Social Neuroscience:
Emotional Perception 2
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI) 1
Image Registration and Computational Anatomy
Multivariate Approaches
Keywords:
Emotions
FUNCTIONAL MRI
Machine Learning
Multivariate
Psychiatric Disorders
Spatial Warping
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:
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.
Chang, L., Jolly, E., Cheong, J. H., Burnashev, A., Chen, A., Frey, S., Clark, M. (2018). NLTools (doi: 10.5281/zenodo.10888639)
Fournier, J. C., Keener, M. T., Mullin, B. C., Hafeman, D. M., LaBarbara, E. J., Stiffler, R. S., Almeida, J., Kronhaus, D. M., Frank, E., & Phillips, M. L. (2012). Heterogeneity of Amygdala Response in Major Depressive Disorder: The Impact of Lifetime Sub-Threshold Mania. Psychological Medicine, 43(2), 293.
Glasser, M. F., Sotiropoulos, S. N., Wilson, J. A., Coalson, T. S., Fischl, B., Andersson, J. L., Xu, J., Jbabdi, S., Webster, M., Polimeni, J. R., Van Essen, D. C., Jenkinson, M., & WU-Minn HCP Consortium. (2013). The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage, 80, 105–124.
Haxby, J. V., Guntupalli, J. S., Nastase, S. A., & Feilong, M. (2020). Hyperalignment: Modeling shared information encoded in idiosyncratic cortical topographies. eLife, 9. https://doi.org/10.7554/eLife.56601
Joshi, A.A., Chong, M., Li, J., Choi, S., Leahy, R.M. (2018). Are you thinking what I’m thinking? Synchronization of resting fMRI time-series across subjects. NeuroImage, 172, 740–752.
Lee, D. S., Sahib, A., Narr, K., Nunez, E., & Joshi, S. (2020). Global Diffeomorphic Phase Alignment of Time-Series from Resting-State fMRI Data. Medical Image Computing and Computer-Assisted Intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention, 12267, 518–527.
Loureiro, J. R. A., Leaver, A., Vasavada, M., Sahib, A. K., Kubicki, A., Joshi, S., Woods, R. P., Wade, B., Congdon, E., Espinoza, R., & Narr, K. L. (2020). Modulation of amygdala reactivity following rapidly acting interventions for major depression. Human Brain Mapping, 41(7), 1699–1710.
Stuhrmann, A., Suslow, T., & Dannlowski, U. (2011). Facial emotion processing in major depression: a systematic review of neuroimaging findings. Biology of Mood & Anxiety Disorders, 1(1), 10.
Weaverdyck, M. E., Lieberman, M. D., & Parkinson, C. (2020). Tools of the Trade Multivoxel pattern analysis in fMRI: a practical introduction for social and affective neuroscientists. Social Cognitive and Affective Neuroscience, 15(4), 487–509.