Automated measurement of intima-media thickness of carotid arteries in ultrasonography by computer software

• Published in: Diagnostic and interventional radiology (Ankara, Turkey) Turkey), 2005-06, Vol.11 (2), p.105-108
• Main Authors: Seçil, Mustafa, Altay, Canan, Gülcü, Aytaç, Ceçe, Hasan, Göktay, Ahmet Yiğit, Dicle, Oğuz
• Format: Article
• Language: English
• Subjects: Adult; Araştırma teknikleri; Arteriosclerosis; Arteriosclerosis - diagnostic imaging; Arteriyoskleroz; Automation; Carotid Artery, Common; Carotid Artery, Common - diagnostic imaging; Diagnosis; Female; Görüntü işleme, bilgisayar destekli; Gözlemci farklılığı; Humans; Image Processing, Computer-Assisted; Investigative techniques; Karotis arter, komon; Male; Observer Variation; Predictive Value of Tests; Software; Tanı; Testlerin kestirim değeri; Tunica Intima; Tunica Intima - diagnostic imaging; Tunika intima; Ultrasonography; Yazılım
• ISSN: 1305-3825; 1305-3612

Intima-media thickness (IMT) has been proposed to be a morphological criterion of atherosclerosis. The purpose of this study was to investigate the interobserver variability of manual and also of computer software measurements of IMT. High-resolution common carotid artery (CCA) images of 88 patients that have been obtained by a linear broadband L5-12 MHz transducer and archived in PACS were retrospectively evaluated. Two separate investigators, who were unaware of the former results, evaluated the same images by using computer software that had a dedicated tool for automatic measurement of IMT. The results of the investigators were compared. According to the two investigators who have performed manual measurements, mean values of IMT of right CCA were 0.6396 mm and 0.6356 mm; of the left CCA were 0.6662 mm and 0.6575 mm, respectively. The interobserver variability of measurements revealed the mean IMT as 0.6071 mm and 0.6048 mm for the right, 0.6216 mm and 0.6227 mm for the left CCA. Manual measurements of both investigators were found to be higher than the automatic measurements and the differences were statistically significant. Interobserver correlation of manual measurements was between 0.80-0.88 and of the automated measurements was between 0.93-0.98. Manual measurements reveal higher values than the automated measurements of IMT. The interobserver correlation of automated measurements is higher than manual measurements. The use of dedicated software may be proposed to reduce the measurement errors.