Clinical evaluation of a fully automated model-based algorithm to calculate left ventricular volumes and ejection fraction using multidetector computed tomography

Eduard Ghersin, Sobhi Abadi, Sergey Yalonetsky, Ahuva Engel, Jonathan Lessick

Research output: Contribution to journalArticle

7 Scopus citations

Abstract

Objectives: To evaluate feasibility and accuracy of a fully automatic algorithm (FAA) for calculating left ventricular volumes and ejection fraction (LVEF) from multidetector computed tomography (MDCT) studies versus a previously validated method based on Simpson's method serving as our reference standard (RS), and left ventriculography (LVG), in patients with acute chest pain (ACP). Methods: 23 patients admitted with ACP underwent ECG-gated MDCT, as well as LVG during their hospitalization. MDCT based end-diastolic, end-systolic and stroke volumes (EDV, ESV, SV) and LVEF were calculated using the RS and the FAA. Results: One patient was excluded after FAA failure to detect contours. In the remaining 22 patients, meanSD LVEF measurements were RS 6111%, FAA 5911% (r = 0.91 versus RS), LVG 5716% (r = 0.6 versus RS). In comparison to RS, FAA overestimated EDV by 10.18.5 ml, ESV by 7.36.4 ml and SV by 3.68.6 ml, and underestimated LVEF measurements by 2.74.6%, related to contour smoothing. LVEF by FAA was within 8% of RS for all patients except one. In contrast, LVG differed significantly from RS. Conclusions: The evaluated FAA obtained accurate, clinically relevant results for left ventricular volumes and LVEF relative to the RS.

Original languageEnglish (US)
Pages (from-to)43-51
Number of pages9
JournalAcute Cardiac Care
Volume11
Issue number1
DOIs
StatePublished - Apr 1 2009

Keywords

  • Acute coronary syndrome
  • Cardiac volume
  • Computed tomography
  • Imaging
  • Left ventricular function

ASJC Scopus subject areas

  • Cardiology and Cardiovascular Medicine
  • Radiology Nuclear Medicine and imaging
  • Emergency Medicine

Fingerprint Dive into the research topics of 'Clinical evaluation of a fully automated model-based algorithm to calculate left ventricular volumes and ejection fraction using multidetector computed tomography'. Together they form a unique fingerprint.

  • Cite this