autohrf-an R package for generating data-informed event models for general linear modeling of task-based fMRI data.

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Biostatistics

Background:

• Task-based functional magnetic resonance imaging (fMRI) commonly uses the General Linear Model (GLM) to analyze brain activity.
• The accuracy of GLM analysis depends on correctly specifying the hemodynamic response function (HRF) and task predictors.
• Inaccurate HRF assumptions can lead to suboptimal models, poor data fit, and invalid brain activity estimates.

Purpose of the Study:

• To develop and evaluate an automated approach for estimating task-specific event models in fMRI GLM analysis.
• To introduce the R package 'AutoHRF' for data-driven estimation and evaluation of HRF parameters.
• To improve the validity and reliability of brain activity estimates in task-based fMRI.

Main Methods:

• Developed the 'AutoHRF' R package utilizing genetic algorithms for automated parameter searching.
• Employed theoretically driven models to define constraints for computationally deriving optimal task response models.
• Validated the 'AutoHRF' package on diverse fMRI datasets, including spatial working memory and flanker tasks, as well as simulated data.

Main Results:

• 'AutoHRF' enables efficient construction and evaluation of improved task-related brain activity models.
• The package optimizes the estimation of task predictor onset and duration, enhancing GLM model fit.
• Demonstrated improved understanding of Blood-BOLD (BOLD) task response and increased validity of model estimates.

Conclusions:

• The 'AutoHRF' package offers a robust solution for data-driven HRF estimation in fMRI.
• Accurate event model specification is critical for sensitive and valid GLM-based fMRI analysis.
• This approach is particularly valuable for complex experimental designs with overlapping events.