Separating function from structure in perfusion imaging of the aging brain

Abstract

The accuracy of cerebral blood flow (CBF) imaging in humans has been impeded by the partial volume effects (PVE), which are a consequence of the limited spatial resolution. Because of brain atrophy, PVE can be particularly problematic in imaging the elderly and can considerably overestimate the CBF difference with the young. The primary goal of this study was to separate the structural decline from the true CBF reduction in elderly. To this end, a PVE-correction algorithm was applied on the CBF images acquired with spin-echo EPI continuous arterial spin labeling MRI (voxel size = 3.4 x 3.4 x 8 mm(3)). Tissue-specific CBF images that were independent of voxels' tissue fractional volume were obtained in elderly (N = 30) and young (N = 26); mean age difference was 43 years. Globally, PVE-corrected gray matter CBF was 88.2 +/- 16.1 and 107.3 +/- 17.5 mL/100 g min(-1) in elderly and young, respectively. The largest PVE contribution was found in the frontal lobe and accounted for an additional 10% and 12% increase in the age-related CBF difference between men and women, respectively. The GM-to-WM CBF ratios were found to be on average 3.5 in elderly and 3.9 in young. Whole brain voxelwise comparisons showed marked CBF decrease in anterior cingulate (bilateral), caudate (bilateral), cingulate gyrus (bilateral), cuneus (left), inferior frontal gyrus (left), insula (left), middle frontal gyrus (left), precuneus (bilateral), prefrontal cortex (bilateral), and superior frontal gyrus (bilateral) in men and amygdala (bilateral), hypothalamus (left), hippocampus (bilateral), and middle frontal gyrus (right) in women.

Figure 1

Group averaged CBF images from HY (A) and HE (B). 1st and 3rd rows show tissue specific images, f _GM and f _WM, respectively. Fractional CBF images, f ${}_{\text{p}}^{\text{GM}}$ and f ${}_{\text{p}}^{\text{WM}}$ for are shown in 2nd and 4th rows, respectively; the last row shows f _NET. Slices were chosen to represent lower, middle, and upper areas of the brain (MNI z‐coordinates: 21, 37, and 41). Units in the bars are in mL/[100 g min⁻¹]. Note good uniformity of the image signal for GM and WM flow densities (1st and 3rd rows).

Figure 2

Pure GM CBF values from PVE‐corrected data (gray) and uncorrected data (black) from HE (A) and HY (B) from ROIs selected to contain voxels with varying GM content. Error bars represent +1 s.e. Note that PVE‐corrected data are independent of ROI's GM content.

Figure 3

Lobewise f _GM (black) and f ${}_{\text{p}}^{\text{GM}}$ (red) for women (A) and men (B). Solid and striped columns represent data from HY and HE, respectively. Error bars represent +1 s.e.

Figure 4

Voxelwise SPM{T} map for the (HY−HE) contrast (corrected at the cluster level of 50 voxels, T > 3.84) overlaid on a surface rendering of the brain for females (A) and males (B) for f _GM (red) and f _NET (green). The overlap of the areas where both f _GM and f _NET were statistically lower in HE are shown in orange.

Figure 5

Histograms of whole brain ratios of f_GM/f_WM for females (A) and males (B) are shown in black and gray for HY and HE, respectively.