Normalisation of ultrasonic images of atherosclerotic plaques and reproducibility of grey scale median using dedicated software

Int Angiol. 2007 Dec;26(4):372-7.

Abstract

Aim: Image normalization using ''Adobe Photoshop'' a commercially available software package designed for photography has provided the means of making reproducible measurements of grey scale in ultrasonic images of carotid bifurcation plaques. A dedicated software package ''Plaque Texture Analysis'' has been developed by the authors that is simple to use, provides facilities for measurement of other texture features in addition to grey scale median (GSM) and allows the operator to save the normalised images, plaques and GSM measurements in a database for subsequent analysis. The aim of the study was to determine (a) the intraobserver variability of GSM when the same plaque images were analysed using both Adobe Photoshop and the dedicated software and (b) the interobserver variability when the same plaque images were analysed using the dedicated software by different observers.

Methods: A sample of 33 images of carotid bifurcation plaques (16 symptomatic and 17 asymptomatic) producing greater than 50% stenosis of the internal carotid artery were analysed by two observers.

Results: The intraclass correlation coefficient and correlation coefficient (r) for GSM were 0.992 (95% CI 0.984 to 0.996) and 0.992 respectively for the first observer (MG); they were 0.986 (95% CI 0.972 to 0.993) and 0.987 respectively for the second observer (AN). The interobserver intraclass correlation coefficient and correlation coefficient (r) were 0.932 (95% CI 0.863 to 0.967) and 0.933 respectively.

Conclusion: These findings and the automatic saving of the normalised images and measurements in a database for subsequent statistical analysis make the ''Plaque Texture Analysis'' software a powerful research tool.

MeSH terms

  • Atherosclerosis / complications
  • Atherosclerosis / diagnostic imaging*
  • Carotid Artery, Internal*
  • Carotid Stenosis / diagnostic imaging*
  • Carotid Stenosis / etiology
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Observer Variation
  • Reproducibility of Results
  • Software*
  • Ultrasonography