Measurement variability of blood-brain barrier permeability using dynamic contrast-enhanced magnetic resonance imaging

Abstract

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is used to quantify the blood-brain barrier (BBB) permeability-surface area product. Serial measurements can indicate changes in BBB health, of interest to the study of normal physiology, neurological disease, and the effect of therapeutics. We performed a scan-rescan study to inform both sample size calculation for future studies and an appropriate reference change value for patient care. The final dataset included 28 healthy individuals (mean age 53.0, 82% female) scanned twice with mean interval 9.9 weeks. DCE-MRI was performed at 3T using a 3D gradient echo sequence with whole brain coverage, T1 mapping using variable flip angles, and a 16-minute dynamic sequence with a 3.2 second time resolution. Segmentation of white and grey matter (WM/GM) was performed using a 3D magnetization-prepared gradient echo image. The influx constant Ki was calculated using the Patlak method. The primary outcome was the within-subject coefficient of variation (CV) of Ki in both WM and GM. Ki values followed biological expectations in relation to known GM/WM differences in cerebral blood volume (CBV) and consequently vascular surface area.

Subject-derived arterial input functions showed marked within-subject variability which were significantly reduced by using a venous input function (CV of area-under-the-curve 46 vs 12%, p < 0.001). Use of the venous input function significantly improved the CV of Ki in both WM (30 vs 59%, p < 0.001) and GM (21 vs 53%, p < 0.001). Further improvement was obtained using motion correction, scaling the venous input function by the artery, and using the median rather than the mean of individual voxel data. The final method gave CV of 27% and 17% in WM and GM respectively. No further improvement was obtained by replacing the subject-derived input function by one standard population input function. CV of Ki was shown to be highly sensitive to dynamic sequence duration, with shorter measurement periods giving marked deterioration especially in WM. In conclusion, measurement variability of 3D brain DCE-MRI is sensitive to analysis method and a large precision improvement is obtained using a venous input function.

More information

Identifiers

ORCID for Aravinthan Varatharaj: orcid.org/0000-0003-1629-5774

ORCID for Ian Galea: orcid.org/0000-0002-1268-5102

Catalogue record

Export record

Altmetrics