Detecting Spatially Consistent Structural Differences in Alzheimer’s and Fronto Temporal Dementia Using Deformation Morphometry

Abstract

Abstract. Atrophy is known to occur at specific sites around the brain in both Alzheimer's disease (AD) and Fronto-Temporal Lobe Dementia (FTLD), inducing characteristic shape changes in brain anatomy. In this paper we employ an entropy driven fine lattice free form registration algorithm to investigate whole brain structural changes induced by these diseases relative to normal anatomy, using deformation morphometry. We focus on Alzheimer's disease (AD) and two common sub groups of FTLD: the frontal lobe variant (FTD) and semantic dementia (SD). The shape of each subject group was characterized at each point in the reference anatomy by the distribution of the determinant of the Jacobian of the transformation mapping the subject to a common reference anatomy. Statistical measures were then applied to locate points where voxel level differences in the Jacobian occur between a control group and each of the disease groups, indicating spatially consistent shape differences induced by a particular disease. Spatial maps of the statistical differences showed very different structural characteristics in each disease. AD was characterized by relative contractions in regions of the hippocampus and the parietal lobe and expansions of the ventricular CSF spaces. FTD was characterized by patterns of contraction in the frontal lobe. SD was characterized by large contractions in the temporal lobe, hippocampus and expansion of the ventricular CSF spaces.

3D pattern of brain atrophy in HIV/AIDS visualized using tensor-based morphometry

“…The parameters of the alignment transformation are tuned to maximize the MI. The MI method has been successfully applied to rigid (Viola and Wells, 1997), non-rigid (Lorenzen et al, 2004;Studholme et al, 2001), and cross-modality registrations (e.g., MRI to PET or histologic images; Kim et al, 1997). Hermosillo (2002) developed a variational formulation to maximize MI using a regularization functional borrowed from linear elasticity theory.…”

“…To save computation time and memory requirements, the source and the target images were filtered with a Hann-windowed sinc kernel, isotropically downsampled by a factor of 2, and registered to a randomly selected control subject's image by maximizing the JRD. As in other TBM studies, e.g., Studholme et al, 2001;Davatzikos et al, 2001, we preferred registration to a typical control image versus a multi-subject average intensity atlas as it had sharper features and, in general, larger effect sizes, as shown in Fig. 1; nevertheless, template optimization for TBM is the subject of further on-going study by us and others (Kochunov et al, , 2005Avants and Gee, 2004;Fletcher et al, 2004;Twining et al, 2005).…”

“…Here we applied tensor-based morphometry (TBM; see Davatzikos et al, 1996;Thompson et al, 2000;Chung et al, 2001Chung et al, , 2003Chung et al, , 2004Fox et al, 2001;Shen and Davatzikos, 2003;Studholme et al, 2001Studholme et al, , 2003Teipel et al, 2004, for related work) to detect and automatically quantify the subtle and distributed patterns of brain atrophy in HIV/AIDS.…”

An Intensity Consistent Approach to the Cross Sectional Analysis of Deformation Tensor Derived Maps of Brain Shape

“…In general we assume that the non-rigid registration transformation between the reference space and each subject T Rn : x R → x n has been estimated by a registration procedure optimising an intensity similarity criteria, C(i R , i n ) between image intensities or alternatively extrapolating transformation models from aligned corresponding features. In cross sectional tensor based morphometry [5,3,6,10], relative differences in shape between the reference and subject anatomies are described by the spatial derivatives of these transformations, evaluated at each point, to form maps of the local point-wise Jacobian,…”

“…counts equivalent intensities falling in bins with width b, indicates nearest integer, and the term Γ = n∈N k∈K f (k) (10) normalises by the weighted total number of voxels. This distribution captures the fraction of the filter neighborhood occupied by correctly aligned intensity pairs, together with the fraction of all combinations of mis-aligned intensity pairs.…”