Ultrasound measurements of carotid intima-media thickness by two semi-automated analysis systems

Increased carotid intima-media thickness (cIMT) is associated with an increased risk of cardiac events and stroke. Several semi-automated edge-detection techniques for measuring cIMT are used for research and in clinical practice. Our aim was to compare two currently available semi-automated techniques for the measurement of cIMT. Carotid ultrasound recordings were obtained from 99 subjects (mean age 54·4 ± 8·9 years, range 33-69) without known cardiovascular diseases using a General Electric (GE) Vivid 7 ultrasound scanner, 8-MHz transducer. The far-wall cIMT was evaluated 1-2 cm proximal to the carotid bulb. Three diastolic images (ECG R-wave) from the left and three images from the right common carotid arteries were analysed using GE and Artery Measurement System (AMS) semi-automated softwares. Mean systolic and diastolic blood pressures were 120 ± 13 and 76 ± 8 mmHg, respectively. The cIMTmean (left + right)/2 by GE and cIMTmean (left + right)/2 AMS were highly correlated (r = 0·92, P<0·001). Higher values were measured by GE (0·72 ± 0·12 mm) compared with AMS (0·69 ± 0·12 mm), and this was significant (P<0·001). The coefficients of variation for the intra-observer variability of cIMTmean (left + right)/2 were 1·0% (GE) and 2·2% (AMS). cIMTmean measured by GE's semi-automated edge-detection method correlated well with that measured by AMS. However, there were small but significant systematic differences between the cIMTmean values measured by the two techniques. Thus, the use of only one type of measurement program seems favourable in follow-up studies and when evaluating treatment effects.