Thresholding lesion overlap difference maps: application to category-related naming and recognition deficits

Area of Science:

• Neuroscience
• Cognitive Psychology
• Neuroimaging

Background:

• Lesion overlap difference maps are used to study brain systems for cognitive tasks like object naming and recognition.
• Previous interpretations focused on maximal overlap, potentially overlooking nuanced anatomical correlations.

Purpose of the Study:

• To develop and apply formal statistical methods for thresholding and power analysis of lesion overlap difference maps.
• To reassess the neural correlates of naming and recognition deficits for concrete entities using improved statistical techniques.

Main Methods:

• Derived exact voxel-wise statistics for lesion overlap and proportion difference maps under a null hypothesis.
• Applied these statistics to a large dataset to analyze lesion correlates of naming and recognition deficits.
• Conducted power analyses to create "effective coverage maps" indicating potential for significant findings.

Main Results:

• Thresholded maps confirmed some previous findings but revealed differences in spatial distributions.
• Confirmed distinct lesion correlates in the inferotemporal region (IT) for naming unique versus nonunique entities.
• Identified the left inferior frontal gyrus (IFG) involvement in naming nonunique natural entities (animals, fruits/vegetables).

Conclusions:

• Formal statistical thresholding and power analysis provide a more robust interpretation of lesion-deficit mapping.
• Effective coverage maps are crucial for interpreting results, especially given lesion coverage heterogeneity.
• Recommends including power maps in voxel-wise lesion-deficit mapping studies using inferential statistics.