Changes in MRI cortical thickness and [18F]FDG PET data over 24 months in subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI) Study

Arnaud Charil, Felix Carbonell, Anthonin Reilhac, Kristina Deduck, Rohit Sood, Alan Evans, Barry Bedell, Alex Zijdenbos
2011 Alzheimer's & Dementia  
There is growing evidence from diffusion tensor imaging (DTI) studies that microstructural integrity can be compromised in MCI. To combine information about macrostructural tissue loss and microstructural degeneration, we conducted a deformation morphometry coanalysis of T1 images with diffusion tensor imaging (DTI). We predicted that co-analysis of structural and DTI would reveal more disease-related brain abnormalities than a voxel-wise analysis of either modality alone. Methods: Twenty-one
more » ... tients with Mild Cognitive Impairment (MCI; 71 6 8 yrs, 11 women, MMSE 29 6 2) and 21 cognitively normal healthy elderly controls (CN; 70 6 7 yrs, 11 women, MMSE 29 6 1) were studied using voxel-wise analysis. Images were registered to standard space, creating maps of deformation (JAC) and fractional anisotropy (FA). Group and age were independent variables in linear models with: (1) FA maps as univariate dependent variables (DV), (2) JAC maps as univariate DVs, and (3) JAC and FA maps as multivariate joint DVs. Results: The top panel of Figure 1 shows analyses (1) and ( 2 ). The left image shows that MCI patients have atrophy in the temporal lobe and posterior cingulate gyrus. The right image shows reduced FA in MCI patients in the posterior corpus callosum. The bottom panel shows results from analysis (3), regions where MCI affected both JAC and FA considered jointly. Figure 2 overlays the significant regions detected in each analysis. Red voxels denote significant regions from the JAC-only analysis, blue voxels from the FAonly analysis, and green from the multivariate analysis. The regions of green in the posterior region of the right corpus callosum and the anterior left temporal lobe are where multivariate analysis was more sensitive than either univariate analysis. Conclusions: The multivariate analysis provides more information than analysis of the individual structural and DTI MRI modalities alone since their covariance is considered, and reveals additional changes in MCI. In particular, it suggests the manifestation of MCI is accompanied by spatially corresponding macro-and microstructural changes in temporal and corpus callosum regions that are not detected by univariate analyses, providing potentially new insight into the mechanisms of brain tissue degeneration.
doi:10.1016/j.jalz.2011.05.583 fatcat:x2rdbp6vzzhpncnxaetrvaelx4