Controlling false positive rates in mass-multivariate tests for electromagnetic responses.

Area of Science:

• Neuroscience
• Biophysics
• Statistical Modeling

Background:

• Accurate analysis of electromagnetic responses in source space is crucial for understanding brain activity.
• Controlling false positive rates in mass-multivariate statistical tests is a significant challenge.
• Existing methods, such as sensor-level multivariate thresholds, can be overly conservative.

Purpose of the Study:

• To develop and validate a more accurate method for controlling false positive rates in mass-multivariate tests.
• To introduce a Bonferroni correction based on unique lead-field extrema for source-level analysis.
• To assess the performance of this new method against empirical permutation thresholds.

Main Methods:

• A novel Bonferroni correction approach using unique lead-field extrema was developed.
• The method was validated using a multivariate beamformer formulation.
• Simulated and real electromagnetic response data were analyzed across various source spaces.
• The approach was tested for both mass-univariate and mass-multivariate tests, including induced and evoked responses.

Main Results:

• The proposed Bonferroni correction is less conservative than traditional sensor-level thresholds.
• The heuristic based on unique lead-field extrema accurately approximates empirical permutation thresholds.
• The method is applicable to both volume and cortical manifold source spaces.
• The framework successfully accommodates multivariate effects beyond single contrasts.

Conclusions:

• The novel Bonferroni correction provides a more efficient and accurate way to control false positive rates in mass-multivariate analyses.
• This method enhances the reliability of interpreting electromagnetic responses in source space.
• The findings support the broader applicability of mass-multivariate statistical frameworks in neuroscience research.