Linear and nonlinear PCA generate equivalent representations of anatomy from AMBA.

(a) First PCA component representation in a sagittal slice. (b) Representation of K-means clustering (K = 200) of PCA components in a coronal slice of the left hemisphere. (c) First KPCA component with Quadratic kernel function in a sagittal slice. (d) Representation of K-means clustering (K = 200) of KPCA components with Quadratic kernel function in a coronal slice of the left hemisphere. (e, f) Adjusted Mutual Information and Adjusted Rand Index of clustered components using K-means clustering with K ranging from 1 to 50 and from 50 to 550 with a step of 50.