Validation of semi-automated flow-mediated dilation measurement in healthy volunteers.

BACKGROUND:Flow-mediated dilation (FMD) is a non-invasive imaging modality used to measure endothelial function but has significant intra- and inter-observer variability. The use of semi-automated FMD devices could overcome this limitation. We assessed the reproducibility of same-day semi-automated FMD measurements by investigators who received basic training on the correct use of the device. METHODS:Forty-three healthy volunteers had two brachial artery FMD measurements performed 20 minutes apart using the UNEX EF 38G device, and automated outputs were produced. Images were also manually analysed using edge-detection software. The reproducibility of repeat FMD measurements within individuals was compared for automated and manual readings, and the correlation between analytical techniques was calculated. RESULTS:Twenty-five percent of scans were of non-diagnostic quality (n = 32). Automated analyses demonstrated sub-optimal reproducibility and measurement variability [intraclass correlation coefficient (ICCC) = 0.334, coefficient of variation (CV) = 45.87%]. In contrast, manually analysed scans had excellent reproducibility and low measurement variance (ICCC = 0.815, CV = 11.40%). FMD values obtained from automated and manual analysis correlated poorly (r = 0.164), whereas resting (r = 0.955) and maximal brachial artery diameters demonstrated excellent correlation (r = 0.867). CONCLUSION:Manually evaluated serial UNEX EF readings have good reproducibility and therefore, the optimal FMD workflow involves manual analyses prior to independent automated interrogation. The high non-diagnostic scan rate is most likely the result of insufficient training and indicates that semi-automatic devices such as UNEX EF should be used by experienced investigators to achieve optimal results.