Improving Brain-Behavior Associations in Functional Connectivity with Selection of Movie Timeframes
Poster No:
1588
Submission Type:
Abstract Submission
Authors:
Sarah Cutts1, Jacob Tanner1, Caio Seguin2, Richard Betzel1, Olaf Sporns1
Institutions:
1Indiana University, Bloomington, IN, 2Department of Psychological and Brain Sciences, Indiana University, Bloomington, IN
First Author:
Co-Author(s):
Introduction:
Past work has shown that identifiability and behavioral signatures in functional connectivity (FC) vary within a scan session [1-4]. This work suggests that multiple levels of variation exist within resting state scans based on the behavior or network property of interest. Movie data has been shown to be promising for assessing FC because it is both more engaging and easier to collect data. It has been shown to enhance behavioral prediction from FC [5] suggested due to similarities in stimuli enhancing notable differences between subjects [6]. It remains unclear whether subject-variability in behavioral measures reveal stronger relationships to specific time points within movies. We optimize the selection of the top 10% of time points within multiple movies that enhance brain-behavioral correlations across 58 separate behaviors.
Methods:
We analyzed fMRI movie data of 126 healthy subjects from the Human Connectome Project [7]. Four scan sessions of movie clips were shown, and each participant performed a battery of out-of-scanner behavioral assessments. Resting-state blocks between movie clips were accounted for by removing 10 TRs at the start of each clip and retaining 5 TRs into resting blocks. An implementation of the Metropolis-Hastings algorithm and simulated annealing [2] was used to find moments within movies that improved correlations with 58 distinct behavioral measures (Fig 1A). Initial timestamps for each movie were randomly generated to select for 80 TRs (~10% of each movie) and these time points were shifted at each step of the optimization to determine which selection of moments improved behavioral correlations. At each step, selected time points were made into functional connectivity components (FCc) [2,8] by computing the agreement matrix across the binarized nodal time series of the selected frames [9]. Magnitude of overall brain-behavior correlations was used as the cost function in the optimization. This was done by computing correlations between each edge of the FCc across subjects with the behavior, then taking the mean absolute value of the correlations across all edges. Selection of time points was then either accepted or rejected based on improvements to the cost function. Timestamps were then altered on 1-3 frames and the process was repeated over 10,000 iterations until the selection of time points converged. One-hundred optimizations were performed on each of the 58 behaviors with separate random selections of 80/20 (101 training / 25 testing subjects) cross validations. Improvements in magnitude of brain-behavior correlations and similarity of brain-behavior correlation edge maps between training and testing groups were assessed by comparing FCc from optimized timestamps to randomly initialized timestamps. Similarities between selected timestamps across behaviors were assessed with Jaccard distance.
Results:
Improvements in brain-behavior correlations and similarity between correlation maps of training and testing groups are shown in Fig 1B for all behaviors of each movie. Multiple behaviors showed significantly (p < 0.05) improved train-test transfer over initialization in the held-out testing subjects and certain behaviors showed significant improvements in brain-behavior correlations or in both measures. Optimizations consistently converged onto similar time points within each of the movies over multiple cross validations (Fig 2A). Notably, the selection of time points and brain-behavior correlation maps differed across behavioral measures (Fig 2B) but shared similar timestamps between some of the behaviors (Fig 2C). Optimized timestamps selected moments that improved variability across subjects (mean standard deviation across edges) for each behavior compared to initial timestamps (Fig 2D).
Conclusions:
Optimizing time point selection in movies to improve behavioral correlations showed consistent convergence onto similar moments within the same behavior and selection of unique moments across unrelated behaviors.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 1
fMRI Connectivity and Network Modeling 2
Methods Development
Keywords:
Computational Neuroscience
Data analysis
FUNCTIONAL MRI
1|2Indicates the priority used for review
Provide references using author date format
[2] Peña-Gómez, C., Avena-Koenigsberger, A., Sepulcre, J., & Sporns, O. (2018). Spatiotemporal network markers of individual variability in the human functional connectome. Cerebral Cortex, 28(8), 2922-2934.
[3] Cutts, S. A., Faskowitz, J., Betzel, R. F., & Sporns, O. (2023). Uncovering individual differences in fine-scale dynamics of functional connectivity. Cerebral Cortex, 33(5), 2375-2394.
[4] Cutts, S. A., Chumin, E. J., Betzel, R. F., & Sporns, O. (2023). Temporal Variability of Brain-Behavior Relationships in Fine-Scale Dynamics of Edge Time Series. bioRxiv, 2023-09.
[5] Finn, E. S., & Bandettini, P. A. (2021). Movie-watching outperforms rest for functional connectivity-based prediction of behavior. NeuroImage, 235, 117963.
[6] Finn, E. S., Scheinost, D., Finn, D. M., Shen, X., Papademetris, X., & Constable, R. T. (2017). Can brain state be manipulated to emphasize individual differences in functional connectivity?. Neuroimage, 160, 140-151.
[7] Van Essen, D. C., Smith, S. M., Barch, D. M., Behrens, T. E., Yacoub, E., Ugurbil, K., & Wu-Minn HCPConsortium. (2013). The WU-Minn human connectome project: an overview. Neuroimage, 80, 62-79.
[8] Chumin, E. J., Cutts, S. A., Risacher, S. L., Apostolova, L. G., Farlow, M. R., McDonald, B. C., ... & Sporns, O. (2023). Edge time series components of functional connectivity and cognitive function in Alzheimer’s disease. Brain Imaging and Behavior, 1-13.
[9] Sporns, O., Faskowitz, J., Teixeira, A. S., Cutts, S. A., & Betzel, R. F. (2021). Dynamic expression of brain functional systems disclosed by fine-scale analysis of edge time series. Network Neuroscience, 5(2), 405–433.
[10] Kong, R., Li, J., Orban, C., Sabuncu, M. R., Liu, H., Schaefer, A., Sun, N., Zuo, X. N., Holmes, A. J., Eickhoff, S. B., & Yeo, B. T. (2019). Spatial topography of individual-specific cortical networks predicts human cognition, personality, and emotion. Cerebral Cortex, 29(6), 2533-2551.