Fig 1.
Methodology and absolute changes in redundancy and synergy after TUS.
A. Twenty two subjects participated in the experiment, N = 22 controls (non-TUS), N = 11 IFC-TUS, and N = 11 Thal-TUS. B. We quantified the redundancy and synergy matrices, computing their median vector across rows and their ranked version, named median-redundancy-rank or median-synergy-rank. C. For each ROI, we compared each median-HOI-rank/absolute distribution (dotted line in B) between the control and TUS. D. We reported the t-values for the absolute changes, representing a shift to an increase (red) or decrease (blue) HOI interactions.
Fig 2.
Relative changes in redundancy and synergy after TUS.
A. Top row: median-redundancy-rank distribution changes after the TUS-IFC. Bottom row: Median-synergy-rank distribution changes after the TUS-IFC. The blue represents a region decreasing the HOI after TUS, wheres the red color describes the increase. B. Similar to A, when the target is the thalamus (TUS-Thal). We reported the t-values corrected by a N = 1000 permutation test in all the comparisons.
Fig 3.
Whole-brain associations between structural models and observed changes in TUS.
For TUS-IFC A. and TUS-Thal B., we computed the models and changes in HOI (after minus before) over a representative matrix (redundancy, synergy, distance, and the communicability, CMY) averaged across all participants. Within each subpanel, each row corresponds to a structural model (distance, top row; communicability model, CMY, bottom row), while each column corresponds to changes in informational quantities (redundancy, left column; synergy, left right column). The darker boxes represent the p-values lower than 0.05 after the Bonferroni correction, with the blue dots representing the redundant and red dots the synergistic changes. The grey colour dots represent the non-significant associations. For the other two models, see S1 Fig.
Fig 4.
Whole-brain modeling predicts the propagation of TUS-induced plasticity from local to global scales.
A. The local dynamics of each node were simulated using the Stuart-Landau oscillator, which, depending on the bifurcation parameter (a), can exhibit sustained oscillators (a>0), noise (a<0) or coexistence of noise-driven and sustained oscillations (a = 0). B. We inform heterogeneous models with communication models based on communicability (CMY) or distance (denoted as ), and a stimulation modulated by the parameter α (denoted by
) for each target. C. Redundancy and synergy fitting between the empirical and simulated data: The x-axis corresponds to different simulated intensities in the model (alpha), while the y-axis to the Spearman correlation between the empirical and simulated statistical differences (all the t-values in TUS minus control). Each column corresponds to the changes in redundancy and synergy for the two targets. Results for the model based on distance are shown with a solid line, those for the one based on the communicability model with a dashed line. D. Corrected t-values in the simulated data (TUS minus control) for the distance-based model (for the communicability model, see S4 Fig). The columns are consistent with panel C. The brain plots illustrate the HOI changes, displaying significant t-values corrected using a permutation test with N = 1000 iterations. Colors indicate negative/positive changes with respect to no stimulation. Coloured triangles represent the stimulated target with significant t-values at negative alpha.