Quantitative MRI of brain fluids

Abstract

Objective: The current gold standard for imaging cerebral Small Vessel Disease (SVD) is to asses structural tissue damage using anatomical MRI. However, the underlying pathology of the blood vessels is not well known and very difficult to visualize. As the brain fluids are in connection to both the blood vessels and the brain tissue, this may offer a window on the brain in health and disease. The overall goal was to explore quantitative MRI of brain fluid properties as a potential measure to study diseases like SVD. Therefore we explored the mapping of perivascular spaces (PVS), cerebrospinal fluid (CSF) production rate using phase-contrast MRI (PC-MRI), and CSF composition differences using T2 mapping of CSF. Methods: An automatic method was developed to annotate PVS in a region-of-interest in the centrum semiovale, using high-resolution 7T scans. Also quantitative measures (length, tortuosity) were extracted. In low-resolution 3T scans manual PVS annotation in a single slice was performed, and texture features were determined. Correlation between the low- and high-resolution PVS counts and between the high-resolution PVS count and the texture features was assessed. Net CSF flow was measured using PC-MRI, with respiratory gating on expiration and on inspiration, and without respiratory gating. For each scan the net CSF flow and stroke volume in the aqueduct over the cardiac cycle was determined. CSF T2-mapping was performed in 7 healthy volunteers at 7T and 3T and was compared with a single echo spin-echo sequence with varying TEs. The influence of partial volume was assessed by analyzing the longest TEs only. B1and B0maps were acquired. B1and B0dependency of the sequences was tested using a phantom. Results: Good correlation was found between the automatically and manually annotated PVS in the high-resolution scans.Poor correlation was found between the low- and high-resolution PVS counts and between the high-resolution PVS count and the texture features. Net CSF flow was caudal during expiration and without respiratory gating, and cranial during inspiration. Respiratory gating did not affect the stroke volume measurements. Repeatability of the net flow was best during inspiration, and poor without respiratory gating. A positive association was found between average stroke volume and net flow difference between inspiration and expiration. At 3T, but not at 7T, peripheral T2,CSFwas significantly shorter than ventricular T2,CSF. Partial volume contributed to this T2difference, but could not fully explain it. B1and B0inhomogeneity only had a limited effect. Conclusions: Automatic PVS annotation is a very promising technique. Extending this to a whole-brain method can further increase it’s value. PVS texture analysis cannot yet be used to determine the PVS load in 3T scans. PC-MRI measurements of CSF proved to be unsuitable as a measure for the CSF production rate, due to the confounding factor of respiration. However, the CSF flow over the cardiac and respiratory cycles may provide a window to map the CSF dynamics. Finally, mapping of intracranial CSF composition (differences) appeared practically unfeasible. However, CSF T2 mapping may give insight in CSF composition differences between groups.