Table 3.

Quantitative comparison of the sensitivity and specificity of the language maps derived from resting‐state and task‐based language fMRIs

Language maps	Sensitivity (number of activated voxels within language ROIs) (mean ± SD)	Specificity (% of activated voxels within language ROIs) (mean ± SD)
RefLC map	1226 ± 471	32.2 ± 8.9
ConLC map	2014 ± 698a	31.1 ± 6.5
task t‐map	1231 ± 611	30.9 ± 12.9

RefLC map: the ICA‐derived “reference language component” map from the resting‐state fMRI; ConLC map: the union map consisting of the activated voxels from all “consensus language components” from resting‐state fMRI; task t‐map: the GLM‐derived t‐map from the task‐based language fMRI.

^aConLC map reveals more voxels within putative language ROIs compared to RefLC map (t = 5.40, P = 0.0001, paired t‐test) and task t‐map (t = 3.42, P = 0.0035).