The functional MRI (fMRI) time-series data from a predefined region of interest (ROI), here the striatum, are rearranged into a time-by-voxels matrix A, as are the time series from all voxels outside the ROI (matrix B). For reasons of computational tractability, the dimensionality of B is losslessly reduced using singular value decomposition (SVD), yielding ∼B. For every voxel within the ROI, its connectivity fingerprint is computed as the Pearson’s correlation (CORR) between the voxel-wise time-series and the SVD-transformed data, yielding matrix C. Then similarity between voxels is computed using the η2 coefficient, resulting in matrix S. Manifold learning using Laplacian eigenmaps is then applied to this matrix, yielding a set of overlapping, but independent, connection topographies or ‘connectivity modes’ that together describe the functional organization of the striatum. These connection topographies indicate how the connectivity profile with the rest of the brain changes across striatum. Voxels that have similar colours in these connectivity modes have similar connectivity patterns with the rest of the brain. Finally, trend surface modelling is applied to summarize the connectivity modes by fitting a set of trend coefficients (β) that optimally combine a set of spatial polynomial basis functions. See Haak et al., 2018 for further details.