Recently, the feasibility of measuring blood flow in the small communicating arteries in the circle of Willis (coW) with 4D flow MRI at 7T was demonstrated¹. The communicating arteries have an important role in redistributing blood to the different parts of the brain. However, it is well known that considerable anatomic variation in the coW exists in the healthy population. In this study, we investigated with 4D flow MRI at 7T if reduced blood flow in hypoplastic or absent cerebral arteries is compensated by increased blood flow in communicating arteries.

Ten healthy volunteers (4 men, 6 women, mean age: 24±2 years, range: 20-27 years) underwent a retrospectively PPU (peripheral pulse unit) gated 4D flow MRI examination on a 7T scanner (Achieva, Philips Healthcare, Cleveland, USA). Non-interpolated spatial resolution was 0.47x0.47x0.50 mm3; TE/TR/FA was 3.1/6.8/20°; VENC was 150 cm/s in the x,y and z-directions. The number of reconstructed cardiac phases ranged from 8 to 13 (mean: 11±2) at a heart rate of 51 to 85 beats per minute (mean: 63±11), resulting in a temporal resolution range of 85-105 ms (mean: 90±7 ms). The scan was accelerated with a SENSE factor of 3 in the right-left direction. Phase images were corrected for concomitant field and eddy current related phase offsets. The lumen of the coW was semi-automatically segmented using commercial software (Mimics, Materialise, Leuven, Belgium). The post-processed images were imported into GTFlow (Gyrotools, Zurich, Switzerland) to visualize pathlines and to quantify blood flow by placing perpendicular planes to the vessels of interest. Flow averaged over the cardiac cycle was measured in the left and right posterior cerebral arteries (L/RPCA), the left and right anterior cerebral arteries (L/RACA), the left and right posterior communicating arteries (L/RPCoA) and the anterior communicating artery (ACoA). Arteries for which flow quantification was not possible due to absence, inability to segment or hypoplasia were excluded from the analysis. Linear regression was used to compare flow values in the LPCA with the LPCoA, the RPCA with the RPCoA, the LACA with the ACoA, the RACA with the ACoA and the LACA with the RACA. The coefficient of correlation (R²) was calculated and P<0.05 was considered a significant correlation.The left and right ACA and the left and right PCoA could be segmented in 9 subjects. The right PCA could be segmented in 8 subjects, whereas the left PCA was segmented in all subjects. The ACoA could be segmented in 9 subjects. In figures 1, 2 and 3, examples are shown of pathlines color-coded for velocity in subjects with a full coW, a coW with a hypoplastic (or absent) RPCA and a coW with a hypoplastic RACA and RPCA. Significant correlations were found for time-averaged flow values in the LPCoA and LPCA, RPCoA and RPCA and ACoA and RACA (figure 4, table 1). Non-significant trends towards flow redistribution were found for ACoA/LACA and LACA/RACA (table 1).Two volunteers had hypoplastic RACAs whereas none had hypoplastic LACAs. Therefore, the ACoA/RACA and LACA/RACA correlations had negative slopes, whereas the ACoA/LACA correlation had a positive slope (more blood flow in the ACoA and LACA related to less flow in the RACA). Such findings will be different if hypoplastic RACAs are present in the cohort, which was not the case. In this small group of young volunteers, 50% of the subjects had coWs with all arteries present without hypoplasia. This number is comparable with numbers reported in the literature². Even at 7T, the image quality of the non-contrast enhanced 4D flow MRI data at a spatial resolution of 0.5 mm does not always allow for segmentation of all arteries. This is partly caused by increased noise in the center of the field of view (where the coW is located) due to SENSE acceleration and the accompanying g-factor increase. More advanced acceleration techniques such as compressed sensing may produce images with higher SNR. However, in most of the subjects, the data quality was sufficient to segment the small arteries to show that flow was redistributed in subjects with hypoplastic arteries.In this study, significant correlations were found for time-averaged flow between the PCoAs and the PCAs and the ACoA and the RACA using 4D flow MRI at 7T. This finding illustrates that flow is redistributed through the communicating arteries in the coW.