A geometric algorithm for medical image correlations.

An automatic image correlation algorithm is discussed for the cranial region which is based on the geometric properties of the second moment tensor associated with a volume. The second moment tensor of two identical volumes represented in two different coordinate frames is evaluated to obtain the set of translations, rotations, and anisotropic scaling operators which relate the two coordinate frames. The theory is presented along with a discussion of quantitative error analysis to measure the usefulness of such an algorithm in the clinical setting.