Results

Figure 3 shows a set of histograms of $S_{\rho }$ for the entire brain volume with the skull and background removed. With no temporal filtering, the histogram has a peak at the low values of $S_{\rho}\approx 15$ representing tissue with high autocorrelation, and a peak at $S_{\rho}=196$ corresponding to white noise.

Figure 3: Histograms of $S_{\rho }$ for a null FMRI dataset with (a) no temporal filtering (b) non-linear high-pass filtering (c) Low pass filtering matched to a Gaussian HRF followed by non-linear high-pass filtering. The only preprocessing is motion correction. Similar results were obtained for five other null datasets from three different subjects on the same scanner. The two apparent populations in (b) are due largely to white and grey-matter; this can be seen in the spatial map of $S_{\rho }$ in figure 4.