Table 1

Descriptive statistics and results of hypothesis tests comparing measurement of aneurysm neck width from standard watershed segmentation, SURGE, and 2D-DSA

Median (95% CI)*
(mm)
Error relative to 2D-DSAvs SURGEvs 2D-DSA
Median (95% CI)*
(mm)
Median (95% CI)*
(%)
W†P valueW†P value
Standard3.83 (3.21 to 4.35)0.49 (0.32 to 0.62)16 (10 to 24)21<0.0001150<0.0001
SURGE3.43 (3.00 to 3.81)0.17 (0.08 to 0.32)6 (3 to 9)4910.002
  • Error relative to 2D-DSA was calculated for standard and SURGE in millimetres (indicated as Median (mm), as in figure 2A), and as a percentage error (indicated as Median (%), as in figure 2B). Statistical differences were tested between paired measurements using the Wilcoxon signed-rank test. The presented errors consider only the width of the neck; measures such as volume, sac diameter, or merging artifacts shown in figure 3A were not quantified.

  • *95% CI generated using percentile bootstrap method with 5000 iterations.

  • †Test statistic for the Wilcoxon signed-rank test.

  • 2D-DSA, two-dimensional digital subtraction angiography; SURGE, segmentation with upsampled resolution and gradient enhancement.