Measurement precision and biological variation of cranial arteries using automated analysis of 3 T magnetic resonance angiography

Background: Non-invasive magnetic resonance angiography (MRA) has facilitated repeated measurements of human cranial arteries in several headache and migraine studies. To ensure comparability across studies the same automated analysis software has been used, but the intra-and interobserver, day-to-day and side-to-side variations have not yet been published. We hypothesised that the observer related, side-to-side, and day-to-day variations would be less than 10%. Methods: Ten female participants were studied using high-resolution MRA on two study days separated by at least one week. Using the automated LKEB-MRA vessel wall analysis software arterial circumferences were measured by blinded observers. Each artery was analysed twice by each of the two different observers. The primary endpoints were to determine the intraclass correlation coefficient (ICC) and intra-an inter-observer, the day-to-day, and side-to-side variations of the circumference of the middle meningeal (MMA) and middle cerebral (MCA) arteries. Results: We found an excellent intra-and interobserver agreement for the MMA (ICC: 0.909-0.987) and for the MCA (ICC: 0.876-0.949). The coefficient of variance within observers was <= 1.8% for MMA and <= 3.1% for MCA; between observers <= 3.4% (MMA) and <= 4.1% (MCA); between days <= 6.0% (MMA) and <= 8.0% (MCA); between sides <= 9.4% (MMA) and <= 6.5% (MCA). Conclusion: The present study demonstrates a low (<5%) inter-and intraobserver variation using the automated LKEB-MRA vessel wall analysis software. Furthermore, the study also suggests that the day-to-day and side-to-side variations of the MMA and MCA circumferences are less than 10%.