Understanding the extent to which vascular disease and its risk factors are associated with prodromal dementia, notably Alzheimer's disease (AD), may enhance predictive accuracy as well as guide early interventions. However, less is known from the imaging perspective about how risk factors such as vascular disease influence AD progression. Here, we apply a new T1 and DTI fusion analysis of the 3D corpus callosum (CC) on mild cognitive impairment (MCI) populations with high and low vascular risk profiles, and compare them to age-matched controls. We hypothesize that the 3D T1 and DTI fusion method will detect new biomarkers that are capable of decoupling the vascular and neurodegenerative components in MCI, and that these may allow us to elucidate the different degenerative patterns that could be used to discriminate MCI subtypes.

Fifty-eight subjects were grouped based on their clinical dementia rating (CDR) and vascular risk profile into 15 MCI subjects with low vascular risk (MCI-l, mean age 76.40 ± 7.65 years), 18 MCI subjects with high vascular risk (MCI-h, mean age 78.39 ± 5.69 years) and 25 healthy controls (mean age 76.68 ± 6.40 years). Brain T1 and DT-MR scans of all the subjects were obtained in a 3T SIEMENS scanner. DTI data were acquired using a b-value of 1000s/mm², and 60 gradient directions. T1 data were acquired using TE=2.98ms, TR=2500ms, and TI=1100ms. All the T1 data were preprocessed and linearly registered to the same template space, selected randomly from one of the controls [1]. On the linearly registered T1 images, each subject's CC was manually traced, and 3D surface representations of the CC were constructed [2]. One-to-one correspondence between vertices were subsequently obtained through constrained harmonic based registration [2]. To obtain correspondence between DTI and T1 information, we linearly transformed the preprocessed DT images from each subject to their corresponding T1 space, and then to the T1 template. After each of the linear registrations, the diffusion tensors were resampled using the b0 transformation matrices, and rotated according to the underlying anatomy [3].

Our statistical analyses were conducted using the following variables:

(1) Morphometry information obtained from surface-based registration: univariate detJ and multivariate (e₁, e₂, e₃) from the logged deformation (Fig.1, 1^st and 2^nd row).

(2) Diffusion information obtained from radius to surface projections: univariate mean FA along the radius of the CC for each vertex and multivariate (λ₁,λ₂) (Fig.1, 3^th and 4^th row). Note: we did not include λ₃. Being small, this value is susceptible to noise and may reduce detection power.

(3) A fusion of morphometry (e₁, e₂, e₃) and diffusion indices (λ₁,λ₂) (Fig.1, 5^th row).

All of the above variables were controlled for age using linear regression. For each of the tests, 10,000 vertex-wise permutations were performed to avoid the normal distribution assumption, and 10,000 structure-wise permutations were used to correct for multiple comparisons [4,5].

In Fig.1, the MCI-l group showed a few alterations scattered on the surface of the CC as compared to controls, while none of the above five measurements detected significant structure-wise differences; the MCI-h group presented broad areas of alterations compared to controls, with significant structure-wise differences detected by (λ1,λ2,e1, e2, e3) measurement. In terms of MCI-h vs. MCI-l, only the fusion measurement reached structure-wise significance, with the main clusters located in the genu of the CC.

Group differences in brain white matter (WM), including the CC, are typically analyzed based on voxels, midlines, or midplanes. However, voxel-based methods give poor localization of differences and may be contaminated by differently oriented tracts [6], while midline- or midplane-based methods rely on assumptions that WM perpendicular to the midline or the midplane is uniformly distributed. Here, we fuse the T1-based morphometry information and DTI-based diffusion information on clearly defined CC regions, that are largely preserved within tract information projected onto the surface of the corpora callosa. In all three group-wise analyses, the fused method successfully outperforms analyses based on structural information or diffusion information alone, showing the feasibility of using the T1 and DTI fusion method to increase detection power.

Moreover, the significant disparities in the genu between MCI-h and MCI-l groups detected by our fusion method are consistent with neuropsychological studies, in which greater impairments in executive function have been reported in MCI with a vascular component [7,8]. Our findings provide further anatomical evidence of vascular contributions to cognitive impairment before clinical manifestations of dementia emerge, which may serve as a new biomarker that helps to elucidate the inter-wired relationship between MCI and vascular risk factors.