Skip to main content
. 2023 Aug 24;128(10):1225–1235. doi: 10.1007/s11547-023-01697-4

Table 2.

Agreement between haemodynamic data obtained with different 4D flow postprocessing software programmes and phase-offset-corrected 2D PC data in 47 patients

Intraclass correlation Bland–Altman analysis Absolute percentage error
ICC 95% CI Bias Range of agreement Median (%) 95% CI
Net flow volume
Software A 0.96* 0.92–0.98 − 3.1 ml (− 4.4%) 24 ml (61%) 10.2 8.9–11.6%
Software B 0.94* 0.92–0.96 − 1.7 ml (− 2.2%) 32 ml (72%) 11.3 10.3–12.8%
Software C 0.96* 0.94–0.97 − 1.2 ml (− 0.8%) 28 ml (70%) 8.9 7.7–10.6%
Software D 0.97* 0.95–0.98 − 2.1 ml (− 3.1%) 23 ml (63%) 10.3 8.6–11.5%
Pulmonary-to-systemic flow ratio (Qp / Qs)
Software A 0.89* 0.82–0.94 0.02 (2.6%) 0.48 (44%) 8.3 7.4–10.3%
Software B 0.88* 0.74–0.94 0.08 (6.7%) 0.47 (45%) 8.3 4.9–11.3%
Software C 0.85* 0.68–0.93 0.07 (7.8%) 0.57 (57%) 9.6 6.8–12.1%
Software D 0.91* 0.83–0.95 0.04 (4.6%) 0.44 (46%) 6.5 4.2–8.9%
Differential blood flow (RPA flow percentage)
Software A 0.33 0.06–0.56 − 1.0% 27% 7.0 5.1–8.9%
Software B 0.35 0.08–0.58 1.4% 29% 7.8 4.7–10.8%
Software C 0.39 0.12–0.60 − 1.6% 29% 6.5 4.0–11.4%
Software D 0.47 0.22–0.67 1.5% 25% 6.0 4.1–11.7%

ICC Intraclass correlation coefficient, CI Confidence interval, SD Standard deviation

*Significantly higher ICC than for 4D flow without phase-offset correction. Software A: Arterys; software B: Circle; software C: Caas; software D: Medis