ESTIMATING CAUSAL EFFECTS IN STUDIES OF HUMAN BRAIN FUNCTION: NEW MODELS, METHODS AND ESTIMANDS

Area of Science:

• Neuroscience
• Statistics
• Medical Imaging

Background:

• Functional magnetic resonance imaging (fMRI) is crucial for studying treatment effects on neural activity.
• Current fMRI analysis often treats individual voxels separately, lacking a causal interpretation framework.
• Existing methods may not fully capture complex brain region interactions or provide robust effect estimations.

Purpose of the Study:

• To define and estimate "point" and "cumulated" causal effects for brain regions using fMRI data.
• To develop a multisubject, multivoxel, multirun whole-brain causal model for enhanced neuroimaging analysis.
• To facilitate inferences about associations between effects in different brain regions.

Main Methods:

• Extended potential outcomes framework to define regional causal effects in fMRI.
• Constructed a whole-brain causal model incorporating regional parameters.
• Utilized averaged blood oxygenation level dependent (BOLD) responses within regions for feasible simultaneous estimation.

Main Results:

• Successfully defined and estimated both point and cumulated causal effects for brain regions.
• Applied the model to a pain study, identifying expected effects in known pain-related brain regions.
• Observed increased cerebellar activity compared to findings from conventional analysis methods.

Conclusions:

• The proposed causal modeling framework offers a more comprehensive approach to analyzing fMRI data.
• The method enables robust estimation of regional causal effects and their interdependencies.
• Findings suggest the cerebellum plays a more significant role in pain processing than previously identified by standard methods.