Organization of intersubject correlations parallels functional gradients during naturalistic viewing
Poster No:
352
Submission Type:
Abstract Submission
Authors:
Meaghan Smith1, Ahmad Samara2, Jeffrey Eilbott3, Hallee Shearer2, Tamara Vanderwal4, Boris Bernhardt5
Institutions:
1McGill University, Montreal, Quebec, 2University of British Columbia, Vancouver, British Columbia, 3BC Children's Hospital Research Institute, Vancouver, British Columbia, 4Department of Psychiatry, University of British Columbia, Vancouver, BC, 5Montreal Neurological Institute and Hospital, Montreal, Quebec
First Author:
Co-Author(s):
Introduction:
Many analyses have been developed to make use of the wealth of data that fMRI produces. Among these methods is gradient analysis, which uses dimensionality reduction to identify a macroscale organizational hierarchy from functional connectivity matrices (1). In contrast, intersubject correlations (ISCs) compare BOLD signal time courses across individuals to describe how reliably a given brain region responds to the same stimulus (2). Despite describing brain organization from different perspectives, both measures appear to capture core features of functional architecture. In particular, ISCs and gradient scores both delineate default mode regions from task-positive brain regions (1, 2, 3). Here, we leverage movie-fMRI to generate robust movie gradients that have previously been shown to enhance brain-behavior associations (4). Movie-watching also synchronizes low-level brain activity which makes higher level individual differences in functional connectivity more identifiable (5). We then use ISC to investigate how intersubject synchronization maps onto the functional hierarchy identified via the gradient analyses. We hypothesize that there will be a significant spatial correlation between gradient scores and ISCs, revealing a parallel organization between BOLD-signal responses to complex stimuli and the hierarchical organization of functional connectivity.
Methods:
Data. These analyses use minimally preprocessed movie-watching data from the Human Connectome Project (HCP) 7T data release (6, 7). From the complete dataset, 95 participants (58 females, mean age 29.5 ± 3.3) from 64 families were selected based on head motion and data availability. One hour of movie data was collected over the course of four 15-minute runs across two sessions. All analyses were first conducted in a discovery dataset of 45 subjects and replicated in the remaining 50 subjects.
Gradient analysis. Gradient analyses were performed parcel-wise using the Schaefer-1000 parcellation (8). The mean time series for each parcel was correlated with the mean time course of all other parcels in the brain to create subject-level FC matrices. Diffusion map embedding was performed at the subject level using the BrainSpace toolbox (9), and individual gradients were aligned to a group-mean template before being averaged to yield group-level gradients.
Intersubject correlations. BOLD-signal timeseries data at the Schaefer-1000 level were used to compute intersubject correlations using a group mean approach. Each subjects' time course at each parcel was correlated with the group average time course at that parcel to provide a single ISC score per region.
Permutation testing. Spatial autocorrelation-preserving null-models (10) were used to assess the significance of correlations between ISCs and scores along the top three gradients. Spin permutation tests were used to generate 10,000 null-models for each comparison.
Gradient analysis. Gradient analyses were performed parcel-wise using the Schaefer-1000 parcellation (8). The mean time series for each parcel was correlated with the mean time course of all other parcels in the brain to create subject-level FC matrices. Diffusion map embedding was performed at the subject level using the BrainSpace toolbox (9), and individual gradients were aligned to a group-mean template before being averaged to yield group-level gradients.
Intersubject correlations. BOLD-signal timeseries data at the Schaefer-1000 level were used to compute intersubject correlations using a group mean approach. Each subjects' time course at each parcel was correlated with the group average time course at that parcel to provide a single ISC score per region.
Permutation testing. Spatial autocorrelation-preserving null-models (10) were used to assess the significance of correlations between ISCs and scores along the top three gradients. Spin permutation tests were used to generate 10,000 null-models for each comparison.
Results:
ISCs and movie gradients followed trends in the literature. ISCs were high in superior temporal and occipital regions. The top three gradients had poles situated in the somatosensory, visual, and auditory cortices respectively and radiated towards heteromodal association systems such as the default mode network. There was a significant spatial correlation between ISC scores and Gradient 2 scores (r = 0.74, p < 0.05), such that regions with the highest and lowest ISC scores were situated at the poles of the second gradient (i.e., visual and default networks). This relationship was reproduced in the replication dataset (r = 0.75, p < 0.05).
Conclusions:
These results provide further support for a macroscale processing hierarchy within the brain that is exemplified under naturalistic conditions. These findings also suggest that when the brain is active and processing complex and ecologically valid stimulus, there is a strong correspondence between functional connectivity patterns at a whole-brain level, and BOLD signal response reliability across subjects at a parcel-level.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism) 1
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling 2
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
ADULTS
Autism
Computational Neuroscience
Data analysis
DISORDERS
FUNCTIONAL MRI
MRI
Perception
Statistical Methods
Other - Gradients
1|2Indicates the priority used for review
Provide references using author date format
(2) Hasson, U. (2004). Intersubject synchronization of cortical activity during natural vision. Science, 303(5664), 1634–1640. https://doi.org/10.1126/science.1089506
(3) Golland, Y. (2006). Extrinsic and intrinsic systems in the posterior cortex of the human brain revealed during natural sensory stimulation. Cerebral Cortex, 17(4), 766–777. https://doi.org/10.1093/cercor/bhk030
(4) Samara, A. (2023). Cortical gradients during naturalistic processing are hierarchical and modality-specific. NeuroImage, 271, 120023. https://doi.org/10.1016/j.neuroimage.2023.120023
(5) Vanderwal, T. (2017). Individual differences in functional connectivity during naturalistic viewing conditions. NeuroImage, 157, 521–530. https://doi.org/10.1016/j.neuroimage.2017.06.027
(6) Glasser, M. F. (2013). The minimal preprocessing pipelines for the Human Connectome Project. NeuroImage, 80, 105–124. https://doi.org/10.1016/j.neuroimage.2013.04.127
(7) Van Essen, D. C. (2013). The Wu-minn human connectome project: An overview. NeuroImage, 80, 62–79. https://doi.org/10.1016/j.neuroimage.2013.05.041
(8) Schaefer, A. (2017). Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI. Cerebral Cortex, 28(9), 3095–3114. https://doi.org/10.1093/cercor/bhx179
(9) Vos de Wael, R. (2020). BrainSpace: A toolbox for the analysis of macroscale gradients in neuroimaging and Connectomics datasets. Communications Biology, 3(1). https://doi.org/10.1038/s42003-020-0794-7
(10) Alexander-Bloch, A. F. (2018). On testing for spatial correspondence between maps of human brain structure and function. NeuroImage, 178, 540–551. https://doi.org/10.1016/j.neuroimage.2018.05.070