2.3.4. Dice coefficients

To quantify alignment between T₁ anatomical and GE EPI data (and thus the effectiveness of distortion compensation), we calculated the overlap between each subject's T₁ and fMRI data in each of the seven analysis conditions using Dice coefficients via AFNI's 3ddot function. The Dice coefficient is a measure of how well two binary datasets overlap in three‐dimensional space. This metric varies between zero (no overlap) and one (identity) and is calculated by taking the intersection of the two datasets, multiplying by two, and then dividing by the total number of voxels in both scans. Dice coefficients were calculated using two different types of binary masks of the 3 and 7 T data: (a) a whole‐brain mask (including some CSF) using AFNI's 3dAutoMask, and (b) a CSF‐excluded mask based on the segmented T₁ and fMRI data (i.e., including white matter [red] and gray matter [blue], but not CSF [green] regions, as shown in Figure 3b,e). Note that the EPI masks were generated from the distortion corrected EPI data themselves, and not from the corresponding field map scans. We also note that both residual distortion and signal loss due to through‐slice dephasing will result in lower Dice coefficient values, indicating poorer agreement with the corresponding T₁ scan.