Assessing the sensitivity of fMRI group maps

Area of Science:

• Neuroimaging
• Statistical analysis
• Brain mapping

Background:

• Group maps summarize brain activation patterns from individual functional maps.
• Various statistical methods exist for combining functional magnetic resonance imaging (fMRI) data.
• Assessing the robustness of these group maps is crucial for reliable interpretation.

Purpose of the Study:

• To evaluate the robustness of different fMRI group map combining procedures.
• To determine the sensitivity of these methods to unusual subject data.
• To examine the theoretical properties and practical performance of four combining methods.

Main Methods:

• Employed the jackknife method to assess sensitivity by iteratively removing subjects.
• Calculated group maps using four different information-combining procedures.
• Quantified robustness using measures comparing full-sample maps with jackknifed maps on fMRI data.

Main Results:

• Identified a trade-off between robustness and conservativeness in combining techniques.
• Liberal methods, detecting more active voxels, were more sensitive to individual subject effects.
• Conservativeness in methods correlated with increased robustness against unusual subjects.

Conclusions:

• The choice of combining method impacts the reliability and generalizability of fMRI group maps.
• Liberal methods offer greater sensitivity but reduced robustness.
• Robustness is essential for accurate representation of group brain activation patterns.