Figure 8: Top: A comparison between corrected P values from 2D GLM (left) and 2D CCA (right), calculated from a random permutation test with 10 000 permutations. The activity maps are thresholded at the same significance level (corrected P = 0.05). The GLM used one isotropic 8 mm FWHM 2D Gaussian smoothing kernel while CCA used one isotropic 2D Gaussian kernel and 3 anisotropic 2D Gaussian kernels, designed such that the largest possible filter that CCA can create has an FWHM of 8 mm. The neurological display convention is used (left is left), 1−p is shown instead of p. Note that CCA detects active voxels in the left motor cortex and in the left somatosensory cortex that are not detected by the GLM. Bottom: A comparison between corrected P values from 3D GLM (left) and 3D CCA (right), calculated from a random permutation test with 10 000 permutations. The activity maps are thresholded at the same significance level (corrected P = 0.05). The GLM used one isotropic 8 mm FWHM 3D Gaussian smoothing kernel while CCA used one isotropic 3D Gaussian kernel and its derivative, designed such that the largest possible filter that CCA can create has a FWHM of 8 mm. The neurological display convention is used (left is left), 1−p is shown instead of p. Note that CCA detects active voxels in the left somatosensory cortex that are not detected by the GLM. 
