P‐043: Identifying the trajectory of disease in APP/PS1 mice using automated deformation‐based analysis with manual validation

Background: A number of recent studies have utilized MRI-based volumetric measures to study the evolution of disease in APP/PS1 transgenic (Tg) mice [1-4]. Objective(s): The objective of this study was to characterize the longitudinal progression of disease in APP/PS1 Tg versus control mice using automated deformation-based analyses of anatomical MRIs in conjunction with established methods using manual volumetry. Methods: T2-weighted MRI scans from 4 time-points, ranging from 2.5 to 9 months of age, were analyzed from 5 APP/PS1 Tg and 5 wild-type mice. MR images were kindly supplied by the MR Center of AstraZeneca R&D Södertälje, Sweden [4]. Each subject scan was nonlinearly registered using a fully-automated, unbiased framework for longitudinal deformation-based analysis [2,5-8]. Trajectories were characterized using basic linear regression. Results: Significant peak voxels at the border of the CA1 field and the dentate gyrus of the hippocampus were found to follow a trajectory similar to that of manually segmented volumes (see figure). As previously described [2], ventricles were found to expand over the course of time in the APP/PS1 Tg mice versus controls. Analysis shows volumetric decreases in the somatosensory cortex and on the ventral side of the thalamus. Conclusions: Nonlinear averages, deformation maps and linear regression maps were automatically generated in a morphometric study of disease progression applied at the voxel-level. Simple linear trajectories were characterized from serial MRI of control and transgenic mice. Longitudinal statistical analysis could be improved by using mixed-effects models allowing for the differentiation of nonlinear trajectories of volume change.