P3‐379: BRAIN DIFFUSION TENSOR IMAGING METRICS IN ISCHEMIC LESIONS IN VASCULAR COGNITIVE IMPAIRMENT

Abstract

There is a strong relationship between cerebrovascular disease and dementia. Statistically, ∼10% of stroke patients have prior history of dementia while ∼10% will develop dementia later. Vascular cognitive impairment (VCI) is a clinical condition that involves cerebrovascular disease accompanied by at least one impaired cognitive domain. A wide range of cerebral tissue injuries indicate the presence of vascular disease including stroke lesions, white matter hyperintensities, and lacunar infarcts. Differentiation of these tissue anomalies is challenging using typical clinical signs and symptoms. Structural neuroimaging, particularly diffusion tensor imaging (DTI) is an MRI technique sensitive to the microstructural alterations of brain tissue. The purpose of this study was to compare water diffusion metrics measured by DTI in different cerebral vascular lesions. As part of the Ontario Neurodegenerative Disease Research Initiative (ONDRI), we examined the variations in DTI metrics of 12 types of cerebral tissues/lesions in subjects with VCI. An automated DTI processing pipeline was developed based on the well-known ENIGMA DTI protocols and freely available image processing software packages. The pipeline incorporated image conversions, quality control and artifact removal, and calculation of the voxelwise diffusion tensors followed by calculation of DTI scalar metrics including fractional anisotropy (FA) and mean diffusivity (MD). Datasets from 20 VCI patients were included in the analysis. Cerebral tissue lesion masks were obtained by semi-automated segmentation of the corresponding T1-weighted images and were used to calculate the average DTI metrics in 12 types of cerebral tissues/lesions. Images from a single VCI subject are provided in Fig. 1. The stroke region (Fig 1b) corresponds to hypo-intense regions in the T1-weighted (Fig 1a) and FA (Fig 1c) images, and hyperintensities in MD (Fig 1d). Figures 2 and 3 provide the average MD and FA variations in different cerebral tissues/lesions in the 20 VCI subjects. FA and MD maps obtained from processing of the brain DTI data of an ONDRI VCI subject. Variations of the mean FA values in diverse cerbral tissues/vascular lesions in 20 ONDRI VCI subjects (error bars represent the standard error of the mean). Cerabral tissue/lesion types include stroke lesions, deep WM hyperintensities (dWMH), periventricular WM hyperintensities (pWMH), deep lacunae (dLACN), periventricular lacunae (pLACN), normal appearing gray matter (NAGM), normal appearing WM (NAWM), left and right hemisphere hippocampal tissues (LSBHV and RSBHV), periventricular spaces (PVS), sulcal CSF (sCSF), and ventricular CSF (vCSF). Variations of the mean MD values in diverse cerbral tissues/vascular lesions in 20 ONDRI VCI subjects (error bars represent the standard error of the mean). Cerebral tissue/lesion types include stroke lesions, deep WM hyperintensities (dWMH), periventricular WM hyperintensities (pWMH), deep lacunae (dLACN), periventricular lacunae (pLACN), normal appearing gray matter (NAGM), normal appearing WM (NAWM), left and right hemisphere hippocampal tissues (LSBHV and RSBHV), periventricular spaces (PVS), sulcal CSF (sCSF), and ventricular CSF (vCSF). Figures 2 and 3 reveal that there are significant variations among the DTI metrics of the cerebral tissues/lesions that may be utilized in the future for automated tissue classification.