Umeå University's logo

A single, arterial-optimized 4D-flow MRI acquisition may enable fast assessment of both cerebral arterial and venous flow. However, arteries and veins require different velocity encoding (VENC) settings for optimal velocity-to-noise ratio (VNR). Consequently, venous measurements using arterial-optimized VENC settings are subject to reduced VNR and require further evaluation. This study compared cerebral venous flow and pulsatility assessments using a high-VENC (110 cm/s, adapted to the arterial system) and a low-VENC (40 cm/s, adapted to the venous system) 4D-flow MRI sequence at 3 Tesla. Flow and pulsatility index (PI) were calculated for cerebral veins, sinuses and internal jugular veins in 36 elderly volunteers (79 ± 5 years). The high-VENC acquisitions allowed visualization of nearly all venous structures. Mean flow differences were small and the correlation, strong, when comparing both acquisitions across sinuses (R = 0.90–0.99, difference = -8–7%) and cortical veins (R = 0.93, difference = − 6%). Inflow-outflow differences at the confluence of sinuses were similar between acquisitions. PI showed moderate to strong agreement except in the straight sinus. Both the vein of Galen and the jugular veins suffered from aliasing in the venous VENC acquisitions. In summary, this study demonstrated that a VENC setting adapted for the arterial cerebral circulation was feasible for studying cerebral venous flow and pulsatility.

Objective: The collapse of the internal jugular vein (IJV) regulates intracranial pressure (ICP) in upright body positions. The cross-section area (CSA) is therefore of interest when studying the effects of postural changes in various neurological diseases. We have developed a semi-automatic segmentation method, which tracks the CSA of the IJV in ultrasound movies, and evaluated its performance in three body positions (supine, 16°, 71°). Approach: The proposed method utilized post-processing image filtering combined with a modified snake active contour algorithm. The ultrasound movies were retrospectively analysed (n = 231, 3s, 28 fps) based on previously collected data from 17 healthy volunteers. The computed CSAs (CA) from the segmentation method were compared to manually segmented CSAs (MA) in two frames per movie. Tracking performance were evaluated by visual inspection. Main results: In the supine position, 100% of the ultrasound movies were tracked successfully, and the mean of CA-MA was −4.4 ± 6.9 mm² (MA, 88.4 ± 50.5 mm²). The most challenging movies occurred in upright body posture where tracking success rate was 90% and mean of CA-MA was −1.4 ± 2.2 mm² (MA, 12.0 ± 11.1 mm²). The semi-automatic segmentations took 55 s to perform on average (per movie) compared to manual segmentations which took 50 min. Significance: Segmentations made by the proposed method were comparable to manual segmentations in all tilt-angles, however much faster. Efficient and accurate tracking of the CSA of the IJV, with respect to postural changes, could help furthering our understanding of how IJV-biomechanics relates to regulation of intracranial pressure in different neurological diseases and physiological states.