Browsing by Author "Banerjee, Soham"
Now showing 1 - 2 of 2
- Results Per Page
- Sort Options
Item Cardiovascular Risk Factors Predict the Spatial Distribution of White Matter Hyperintensities(2014-02-04) Banerjee, Soham; King, Kevin; McColl, Roderick; Whittemore, Anthony; Hulsey, Keith; Peshock, Ronald M.PURPOSE: Increased volume of brain white matter hyperintensities (WMH) seen on MRI is associated with cardiovascular risk factors; however, WMH have also been attributed to normal aging. Recent studies have suggested that WMH in some brain regions are more strongly associated with specific risk factors. The purpose of this study was to create a map of every individual brain voxel that was significantly associated with risk factors (hypertension, diabetes, hyper-cholesterolemia) as compared to those without each risk factor. The aim of the study is to create a predictive model, which uses the WMH distribution to determine the associated underlying risk factor. METHODS: The MRI brain images used for analysis were obtained from 2066 participants in the Dallas Heart Study, a population based study. Each MRI brain was transformed onto a standard template that adjusts for participant variation in brain volume and shape, using the FSL SIENAX software. The participant's WMH distributions were then generated from their MRIs using an automated algorithm. For each risk factor, the subjects were divided into a case group and a control group. Each voxel of WMH was compared between the two groups using a two tailed nonparametric permutation test. A map of every voxel significantly associated with each risk factor was created. RESULTS: Of the total of 431891 voxels that comprise the distribution of WMH over the entire population, 26064 voxels (6%) were significantly associated with hypertension only. These hypertensive-associated voxels were prevalent anterior to the frontal horns of the lateral ventricles. Similarly, 22527 voxels (5%) were associated with diabetes only with a prevalence near the longitudinal cerebral fissure as well as lateral to the posterior horns of the lateral ventricles. 8088 voxels (2%) were associated with hyper-cholesterolemia only and were abundant posterior to the posterior horns of the lateral ventricles. 331588 voxels (77%) were not associated with a risk factor. CONCLUSIONS: For hypertension, diabetes, and hyper-cholesterolemia, certain voxels were significantly associated with a risk factor, and maps of these voxels were created. Knowing the WMH distribution significantly associated with each risk factor will improve the specificity for evaluating patients for risk factor associated white matter injury. Importantly, this approach makes no a priori assumptions which divide the brain into functional regions or vascular territories.Item Cardiovascular Risk Factors Predict the Spatial Distribution of White Matter Hyperintensity(2015-03-24) Banerjee, Soham; McColl, Roderick W.; Whittemore, Anthony W.; Hulsey, Keith M.OBJECTIVES: To identify the different spatial distribution of white matter hyperintensity (WMH) associated with specific risk factors and use this distribution to estimate the extent of risk factor associated WMH in an individual. MATERIALS AND METHODS: MRI brain images were obtained from 2066 healthy adult participants (858 males, 1208 females; mean age: 50) from a population based sample. An automated algorithm generated each participant’s WMH distribution, registered onto the MNI-152 standard template. For univariate analysis, each risk factor group was compared to the non-risk factor group. Voxels in which WMH frequency was significantly higher (p<0.05) in the risk factor group were mapped. Multivariate analysis consisted of subgroup analysis to minimize confounding of a risk factor on the others. RESULTS: 431891 MNI-space voxels comprised WMH distribution of the entire population. For univariate analysis, 23697 voxels (5.5%) of these voxels were exclusively associated with hypertension and were prevalent in the anterior frontal lobe. Similarly, 24637 voxels (5.7%) were exclusively associated with diabetes and were prevalent at the callososeptal interface. 7315 voxels (1.7%) were only associated with hypercholesterolemia and did not form a discrete spatial distribution. 282115 voxels (65.3%) were not associated with any of the specified risk factors. Multivariate results corroborated the univariate findings. CONCLUSIONS: Each risk factor was associated with a different spatial distribution of WMH. Hypertension was associated with WMH in the anterior frontal lobe and diabetes was associated with WMH in the callososeptal interface.