Skip to main content
. 2011 Jan 19;100(2):440–449. doi: 10.1016/j.bpj.2010.10.053

Figure 2.

Figure 2

Clusters of voxels at which the TDED exceeds 5σ, obtained by applying the k-means clustering algorithm with the Pearson squared similarity measure to simulated data. Each panel represents a different cluster. The number of voxels in each cluster is shown in parentheses: (a) no added noise: A(54), B(44), C(51), D(50), E(32), F(39), G(36), H(37), I(36), J(43), K(18), L(65), (b) intermediate (2 s/1 s) noise level: A(40), B(33), C(21), D(14), E(10), F(48), G(2), H(38), I(12), J(21), K(33), L(15), and (c) high (5 s/3 s) noise level: A(5), B(14), C(6), D(27), E(22), F(22), G(20).