Graph learning for functional brain connectivity: An empathy network study

Abstract

Functional Magnetic Resonance Imaging (fMRI) research employing naturalistic stimuli, particularly movies, examines brain network interactions underlying complex cognitive processes such as empathy. Leveraging graph learning methods applied to whole-brain time-series signals, a novel processing pipeline is proposed, integrating high-pass filtering, voxel-level clustering, and windowed graph learning with a sparsity-based approach. The study involves the analysis of fMRI data collected from healthy participants while watching empathetic movies and during resting-state conditions. Key brain regions implicated in the empathy network, including the Insula, Pre-Frontal Cortex (PFC), Anterior Cingulate Cortex (ACC), and parietal regions, are examined. Results of the exploratory analysis reveal that the sparsity-based graph learning method consistently outperforms others in capturing graph cluster label variations in comparison with the emotion contagion scale, achieving over 88% match across participants. The analysis demonstrates a gradual induction of empathy with a match after 150 seconds through the stimulus. Additionally, edge-weight dynamics analysis of the edge between empathy supporting areas underscores the superiority of sparsity-based learning, with some providing noisy activations. Connectome-network analysis highlights the pivotal role of the Insula, Amygdala, and Thalamus in empathy, with lateral brain connections facilitating synchronized responses. Spectral filtering analysis emphasizes the significance of the band-pass filter in isolating regions linked to emotional and empathetic processing during high emotional states. Strong similarities across movies in graph cluster labels, connectome-network analysis, and spectral filteringbased analyses reveal robust neural correlates of empathy. Furthermore, a comparative study of task and resting-state conditions reveals alignment with the resting-state during low emotional valence intervals of the movie but diverges notably during high emotional valence intervals, suggesting a shared connectivity pattern between stimulus-induced directed (controlled) mind wandering (bottom-up process) and resting-state activity (top-down process). The sparsity-based method shows a 98% match with viewer ratings on the emotion contagion scale, surpassing the 84% match achieved by Pearson’s correlation-based method. This nuanced understanding of neural dynamics in empathy-related tasks versus resting-state enhances the understanding of the networks underlying cognitive processes in real-life situations (naturalistic) and the use of resting-state for the same, paving way for targeted interventions and treatments for conditions associated with empathetic processing and offering significant real-life applications and impact.