Automated edge-detection technique for measurement of brachial artery reactivity: a comparison of concordance with manual measurements Academic Article uri icon

abstract

  • Concerns have been raised about the reproducibility of brachial artery reactivity (BAR), because subjective decisions regarding the location of interfaces may influence the measurement of very small changes in lumen diameter. We studied 120 consecutive patients with BAR to address if an automated technique could be applied, and if experience influenced reproducibility between two observers, one experienced and one inexperienced. Digital cineloops were measured automatically, using software that measures the leading edge of the endothelium and tracks this in sequential frames and also manually, where a set of three point-to-point measurements were averaged. There was a high correlation between automated and manual techniques for both observers, although less variability was present with expert readers. The limits of agreement overall for interobserver concordance were 0.13 +/- 0.65 mm for the manual and 0.03 +/- 0.74 mm for the automated measurement. For intraobserver concordance, the limits of agreement were - 0.07 +/- 0.38 mm for observer 1 and - 0.16 +/- 0.55 mm for observer 2. We concluded that BAR measurements were highly concordant between observers, although more concordant using the automated method, and that experience does affect concordance. Care must be taken to ensure that the same segments are measured between observers and serially.

authors

publication date

  • September 2001