guest ::
| Home > Publications database > Estimating task structure by maximizing similarity between task-state and resting-state functional connectivity |
| Home > Publications database > Estimating task structure by maximizing similarity between task-state and resting-state functional connectivity |
| Poster (After Call) | FZJ-2026-03472 |
; ; ; ; ; ; ; ;
2026
This record in other databases:
Please use a persistent id in citations: doi:10.34734/FZJ-2026-03472
Abstract: The brain flexibly reconfigures in response to task demands. Task designs induce time-locked changes in the blood-oxygen-level-dependent (BOLD) response. Removing variance of the task’s structure yields strong associations between task-removed andresting-state functional connectivity (trFC, rsFC) (Cole et al. 2015). Capitalizing on this, we investigated whether the task structure can be identified in a data-driven way. Data from the first and second phase-encoding runs of 100 subjects from the HumanConnectome Project (HCP-YA) were analyzed separately. Spearman correlation between trFC and rsFC was used as the optimization target for genetic algorithm (GA) to estimate task onset times of a working memory paradigm. The GA-derived optimal task structure was used as task regressor to estimate the explained BOLD signal variance and was compared to results from the known task structure. GA-derived trFC’s correlation to rsFC surpassed the correlation derived from the baseline of known structure removal to rsFC. The GA-derived task onsets aligned well with known onset times in the first run. For the second run, GA-derived onsets were not temporally aligned with known onsets. Task activation analyses using GA-derived task structures identified clusters comparable to those derived from the known task structure in both runs. These findings suggest that maximizing the similarity between task-removed andresting-state FC can decipher task structure. Precise temporal alignment with ground truth may not be necessary to recover meaningful activation patterns, possibly reflecting sensitivity to task-related network reconfigurations that are not time-locked to the task onset.
|
The record appears in these collections: |
PDF