Low dimensional transformations focusing on whole-brain alignment such as Talairach techniques are significantly limited in their ability to account for variability, shape, and location of brain structures. CIS researchers have developed high dimensional large deformation diffeomorphic metric mapping (LDDMM) for the alignment of Medial Temporal Lobe (MTL) structures and their functional responses to common extrinsic atlas coordinates. Increasing the dimension of the mappings for structural alignment with LDDMM increases the accuracy of data activity localization. Also this technique is general and can be applied throughout the brain.

The structural data consists of gray-scale MRI images of 20 subjects whose fMRI data were recorded during declarative memory tasks. Rough segmentations of MTL structures are used in LDDMM to align the fMRI data of these subjects into common atlas coordinates. The improved alignment improves the ability to localize data activity and this assigns functional activity to correct gray matter regions and not to white matter regions. The rows of the top image show sections of enhanced functional responses (color) in the MTL. The statistical analysis of this functional data is shown in the lower left graph. These figures show that the functional data overlap is increased as the methods for registration for functional data are refined from Talairach to LDDMM.