Effects of the Wishart rolloff on dense functional connectivity maps of both an individual subject and group data (210 HCP subjects; MSMAll surface registration). Top rows show an individual subject, before (column 1) and after (column 2) Wishart rolloff for a seed location in lateral parietal cortex (white dot in upper left panels). The correlation increases dramatically as unstructured spatio-temporal noise is reduced, however the map is not substantially "smoothed" as it would be with typical smoothing algorithms. Bottom rows show a group dataset before and after Wishart rolloff for a seed location in the posterior cingulate sulcus (white dot in lower left panels). The dataset has been created using the MIGP algorithm to generate a group PCA series (d=4500) that represents the group concatenated timeseries. Because of the hard cutoff at PCA component number 4500, there is a 'ringing' pattern resulting from spatial autocorrelation in the spatio-temporal noise that is represented by the PCA components with the lowest eigenvalues. If a Gaussian filter had been applied, this pattern of "local connectivity" would be a blob instead of rings. The Wishart rolloff eliminates these rings and again dramatically increases the SNR of the data.
Surface Mesh:32k fs LR, Registration:MSMAll, Species:Human, Modality:Myelin Map, Modality:T2-weighted, Modality:T1-weighted