Characterization and correction of beam-hardening artifacts during dynamic volume CT assessment of myocardial perfusion

Kakuya Kitagawa, Richard T. George, Armin Arbab-Zadeh, João A.C. Lima, Albert C. Lardo

Research output: Contribution to journalArticle

Abstract

Purpose: To fully characterize beam-hardening effects caused by iodinated contrast medium in the left ventricular (LV) cavity and aorta in the assessment of myocardial perfusion at computed tomography (CT) and to validate a beam-hardening artifact correction algorithm that considers fluid-filled vessels and chambers important sources of beam hardening. Materials and Methods: The Johns Hopkins University animal care and use committee approved all procedures. An anatomically correct LV and myocardial phantom to characterize beam-hardening artifacts was designed. Following validation in the phantom, the beam-hardening correction(BHC) algorithm was applied to 256-detector row dynamic volume CT images in a canine ischemia model (n = 5) during adenosine stress, and the effect of beam hardening was determined by comparing regional dynamic volume CT perfusion metrics (myocardial upslope normalized by maximum LV blood pool attenuation) with microsphere-derived myocardial blood flow (MBF). A paired Student t test was used to compare continuous variables from the same subject but under different conditions, while linear regression analysis was performed to estimate the slope and statistical signifi cance of the relationship between CT-derived perfusion metrics and microsphere-derived MBF. Results: Beam-hardening artifacts were successfully reproduced in phantom studies and were eliminated with the BHC algorithm. The correlation coefficient of CT-derived perfusion metrics and microsphere-derived MBF improved from 0.60 to 0.74 (P > .05) following correction in the animal model. Conclusion: Beam-hardening artifacts confound dynamic volume CT assessment of myocardial perfusion. Application of the BHC algorithm is helpful for improving accuracy of myocardial perfusion at dynamic volume CT.

Original languageEnglish (US)
Pages (from-to)111-118
Number of pages8
JournalRADIOLOGY
Volume256
Issue number1
DOIs
StatePublished - Jul 1 2010

    Fingerprint

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this