Figure 11.

Classifying rat fMRI BOLD data. Signal classification was performed on the 11.7T BOLD dataset shown in Figure 10 by the PSD and SSC methods (A) previous tested in this capacity by Eke et al. (2002); misclassification rates for PSD and SSC are shown in the plots of (B) the lower panel. Because SSC is a much better classification tool, than PSD is, the classification topology will be drastically different for these two methods. The ROI’s corresponding to voxel-wise signals identified by SSC as non-stationary indeed do clearly delineate the anatomical boundaries of the brain cortex, while those by the PSD only the spots of highest β.