Comparison of phantom and computer‐simulated MR images of flow in a convergent geometry: Implications for improved two‐dimensional MR angiography

John M. Siegel, John N. Oshinski, Roderic I. Pettigrew, David N. Ku

Research output: Contribution to journalArticle

18 Scopus citations

Abstract

The signal loss that occurs in regions of disturbed flow significantly decreases the clinical usefulness of MR angiography in the imaging of diseased arteries. This signal loss is most often attributed to turbulent flow; but on a typical MR angiogram, the signal is lost in the nonturbulent upstream region of the stenosis as well as in the turbulent downstream region. In the current study we used a flow phantom with a forward‐facing step geometry to model the upstream region. The flow upstream of the step was convergent, which created high levels of convective acceleration. This region of the flow field contributes to signal loss at the constriction, leading to overestimation of the area of stenosis reduction. A computer program was designed to simulate the image artifacts that would be caused by this geometry in two‐dimensional time‐of‐flight MR angiography. Simulated images were compared with actual phantom images and the flow artifacts were highly correlated. The computer simulation was then used to test the effects of different orders of motion compensation and of fewer pixels per diameter, as would be present in MR angiograms of small arteries. The results indicated that the computational simulation of flow artifacts upstream of the stenosis provides an important tool in the design of optimal imaging sequences for the reduction of signal loss.

Original languageEnglish (US)
Pages (from-to)677-683
Number of pages7
JournalJournal of Magnetic Resonance Imaging
Volume5
Issue number6
DOIs
StatePublished - 1995

Keywords

  • Angiogram
  • Computer simulation
  • Convective acceleration
  • Signal loss

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Fingerprint Dive into the research topics of 'Comparison of phantom and computer‐simulated MR images of flow in a convergent geometry: Implications for improved two‐dimensional MR angiography'. Together they form a unique fingerprint.

Cite this