Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (2 results)

  • Thesis
  • Peer Reviewed
UC Riverside Electronic Theses and Dissertations (2024)

This dissertation addresses critical challenges in causal discovery from whole-brainfMRI data, enhancing scalability, capturing contemporaneous effects, and handling complex neural dynamics. Popular methods like Granger Causality (GC) and Dynamic Causal Modeling (DCM) are limited by their scalability and inability to handle instantaneous effects and latent confounds, motivating a new approach. First, a comparative analysis reveals significant limitations in existing methods when applied to whole-brain fMRI data, emphasizing the need for improved accuracy and scalability. In response, we introduce CaLLTiF (Causal Discovery for Large-scale Low-resolution Time-series with Feedback), a novel constraint-based method that leverages conditional independence tests on both lagged and contemporaneous interactions.

Validated on simulated fMRI data from the macaque connectome, CaLLTiF achievessuperior accuracy and scalability compared to existing methods. Applied to resting-state human fMRI, CaLLTiF constructs causal connectomes that exhibit consistent top-down influence from attention and default mode networks to sensorimotor areas, show Euclidean distance-dependent interactions, and are largely dominated by contemporaneous effects. These findings establish CaLLTiF as a powerful tool for capturing large-scale brain interactions in resting states. Further, CaLLTiF is applied to Decoded Neurofeedback (DecNef) studies, a technique combining real-time fMRI with multivariate pattern analysis. In this meta-study of five DecNef experiments, CaLLTiF reveals differences in causal connectivity between neurofeedback and control sessions, with certain connections positively correlated with neurofeedback performance. Neurofeedback graphs display unique dynamics, such as reduced average shortest path lengths and higher right-limbic nodal degree. Additionally, connections within bilateral control networks correlated with neurofeedback scores, underscoring key brain areas involved in neurofeedback efficacy. Distinctions in somatomotor connectivity further separate cognitive-focused from perception-focused neurofeedback sessions. Overall, this dissertation introduces CaLLTiF as a scalable and accurate solution for whole-brain causal discovery, offering valuable insights into resting-state brain organization and task-specific dynamics in neurofeedback. These findings provide new foundations for exploring causal brain interactions in both health and clinical neuroscience.

    Cover page: Mapping Whole-Brain Causal Connectome Using fMRICreative Commons 'BY-NC' version 4.0 license
    • Article
    • Peer Reviewed
    UC Irvine Previously Published Works (2025)
    Despite significant research, discovering causal relationships from fMRI remains a challenge. Popular methods such as Granger causality and dynamic causal modeling fall short in handling contemporaneous effects and latent common causes. Methods from causal structure learning literature can address these limitations but often scale poorly with network size and need acyclicity. In this study, we first provide a taxonomy of existing methods and compare their accuracy and efficiency on simulated fMRI from simple topologies. This analysis demonstrates a pressing need for more accurate and scalable methods, motivating the design of Causal discovery for Large-scale Low-resolution Time-series with Feedback (CaLLTiF). CaLLTiF is a constraint-based method that uses conditional independence between contemporaneous and lagged variables to extract causal relationships. On simulated fMRI from the macaque connectome, CaLLTiF achieves significantly higher accuracy and scalability than all tested alternatives. From resting-state human fMRI, CaLLTiF learns causal connectomes that are highly consistent across individuals, show clear top-down flow of causal effect from attention and default mode to sensorimotor networks, exhibit Euclidean distance dependence in causal interactions, and are highly dominated by contemporaneous effects. Overall, this work takes a major step in enhancing causal discovery from whole-brain fMRI and defines a new standard for future investigations.
      Cover page: Whole-brain causal discovery using fMRI.Creative Commons 'BY' version 4.0 license
      Top