Towards whole brain mapping of the hemodynamic response function.

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Biophysics

Background:

• Functional magnetic resonance imaging (fMRI) conventionally uses a linear model with a stereotyped hemodynamic response function (HRF) to analyze brain activity.
• The neural signal and resulting blood-oxygen-level-dependent (BOLD) response can vary based on task, brain region, and individual subject characteristics.
• The shape of the HRF itself may contain valuable physiological information.

Purpose of the Study:

• To investigate the variability of the BOLD signal's hemodynamic response shape across different tasks, brain regions, and subjects.
• To determine if a task-specific deconvolved HRF (dHRF) offers a more accurate representation than a canonical HRF.
• To assess the implications of HRF variability on fMRI analysis models and statistical outcomes.

Main Methods:

• BOLD signal data from various sensory and cognitive tasks were analyzed.
• A sine series expansion was used to fit the BOLD signal.
• Modeled signals were deconvolved to generate task-specific dHRFs, characterized by amplitude, latency, time-to-peak, and full-width at half maximum.

Main Results:

• The BOLD response shape exhibited significant variability across activated regions, tasks, and even among subjects.
• Amplitude and latency showed the largest variations, while time-to-peak and full-width at half maximum were more consistent.
• The derived dHRFs frequently deviated from the canonical HRF shape in several brain regions.

Conclusions:

• A single, standard HRF may not be optimal for all fMRI analyses, potentially leading to model misspecification.
• The observed variability in HRF shape underscores the need for more personalized or task-specific modeling approaches in fMRI.
• Failure to account for HRF variability could introduce statistical bias in fMRI results.