Coronary artery calcium scoring: diagnostic accuracy of different software implementations

Invest Radiol. 2003 Dec;38(12):761-8. doi: 10.1097/01.rli.0000084257.43062.c2.

Abstract

Purpose: To compare the diagnostic power of different software implementations for the quantification of coronary artery calcium.

Materials and methods: Electron beam computed tomography was performed in 109 heart transplant recipients at the same time as catheter coronary angiography and intracoronary ultrasound. Electron beam computed tomography images were analyzed by 3 software packages marketed for the quantification of coronary calcifications using the same software settings, and the resultant calcium scores correlated with the invasive reference methods by Bland-Altman plots and analysis of the receiver operating characteristics.

Results: Although all scoring systems displayed close correlations upon regression analysis (r2=0.94-0.99), their ability to detect disease as per the invasive reference method varied significantly in some instances. The area under the ROC curve varied between Az=0.78 and 0.85 for the detection of coronary artery stenosis upon coronary angiography (P=0.05-0.13), and between Az=0.75 and 0.83 for the detection of accelerated intimal proliferation (P=0.03-0.18).

Conclusions: Different software implementations for the quantification of coronary artery calcium load may display diagnostically relevant differences in spite of close direct correlation.

Publication types

  • Comparative Study

MeSH terms

  • Adolescent
  • Adult
  • Aged
  • Algorithms
  • Calcinosis / diagnostic imaging*
  • Coronary Artery Disease / diagnostic imaging*
  • Female
  • Humans
  • Male
  • Middle Aged
  • ROC Curve
  • Sensitivity and Specificity
  • Software Validation*
  • Tomography, X-Ray Computed*