Sample size estimation for comparing parameters using dynamic causal modeling.

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Computational Psychiatry

Background:

• Functional magnetic resonance imaging (fMRI) is vital for studying brain activation in response to illness and pharmacological interventions.
• Effective connectivity analyses are increasingly used in fMRI to understand task-related neural networks.
• Determining appropriate sample sizes is essential for the reliable application of these analytical techniques.

Purpose of the Study:

• To present a novel method for estimating the required sample size in fMRI studies.
• To ensure adequate statistical power for effective connectivity analyses.
• To provide a practical guide for researchers planning fMRI studies.

Main Methods:

• Utilized Bayesian Model Selection to identify the optimal statistical model.
• Employed a bootstrapping technique to estimate parameter variance.
• Applied the developed method to two distinct fMRI tasks for validation.

Main Results:

• The proposed method provides a robust estimation of sample size requirements.
• For the illustrative tasks, approximately 20 volunteers per group were found to be sufficient.
• Results highlight the need to evaluate sample size requirements on a case-by-case basis due to data variability.

Conclusions:

• The developed Bayesian approach offers a reliable method for sample size estimation in fMRI.
• This technique is particularly useful for studies employing Dynamic Causal Modeling.
• Researchers should tailor sample size calculations to specific experimental designs and datasets.