Automated vessel edge detection in velocity-encoded cine-MR (VEC-MR) flow measurements: A retrospective evaluation in critically ill patients

Christine B. Henk, Stephan Grampp, Werner Backfrieder, Jasmin Liskutin, Christian Czerny, Gerhard H. Mostbeck

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)


Objective: To assess feasibility of automated edge detection in magnetic resonance (MR) flow calculations in a clinical setting with critically ill patients. Material and methods: Velocity encoded cine-MR (VEC-MR) flow measurements cross-sectional area (CSA), mean spatial velocity (MSV), instantaneous flow (IF), flow (F), 0.5 T Philips, TR 800-800, TE=8 ms, 30° flip angle, FOV 280 mm, 128×256 matrix, temporal resolution 16 time frames/RR, VENC=120 cm/s) were obtained in 20 major thoracic human vessels (ascending aorta, main, right and left pulmonary artery - AAO, MPA, RPA, LPA) of five patients, suffering from severe chronic thromboembolic pulmonary hypertension (CTEPH). Flow maps were evaluated by two independent observers using conventional manual edge detection (INTER m/m). Flow calculations were performed by one observer using both, manual and automated edge detection (INTRA m/a), by a second observer using automated edge detection two times (INTRA a/a) and by two independent observers using automated edge detection (INTER a/a). Evaluation time was measured. Linear regression analysis and Student's t-test were performed. Results: Overall regression coefficients (r2) for INTER m/m, INTRA m/a, INTER a/a and INTRA a/a, respectively, were as follows: CSA, 0.91, 0.91, 0.96, 0.98; MSV, 0.97, 0.99, 0.99, 0.99; IF, 0.98, 0.99, 0.99, 0.99; F, 0.98, 0.99, 0.99, 0.99. Manual CSA values differed significantly from automated data in MPA (P=0.01), RPA (P=0.0008) and LPA (P=0.02). No difference was found for the other assessed parameters of the pulmonary circulation. Average evaluation time per vessel was 20.2±2.6 min for manual and 2.1±0.7 min for automated edge detection (P<0.00001). Conclusion: The software program used provided reproducible data, lead to a 90% reduction in evaluation and calculation time and, therefore, might excel the utilization of VEC-MR flow measurements. Despite variations in the evaluation of the pulmonary circulation CSAs, flow assessment is feasible in critically ill patients.

Original languageEnglish
Pages (from-to)274-281
Number of pages8
JournalEuropean Journal of Radiology
Issue number3
Publication statusPublished - Dec 2003
Externally publishedYes


  • Computers
  • Diagnostic aid
  • Flow dynamics
  • Magnetic resonance (MR)
  • Pulmonary arteries
  • Vascular studies
  • Blood Flow Velocity/physiology
  • Humans
  • Magnetic Resonance Angiography/methods
  • Middle Aged
  • Critical Illness
  • Thoracic Arteries/pathology
  • Linear Models
  • Male
  • Magnetic Resonance Imaging, Cine/methods
  • Feasibility Studies
  • Hypertension, Pulmonary/diagnosis
  • Female
  • Retrospective Studies
  • Chronic Disease
  • Observer Variation
  • Pulmonary Circulation/physiology


Dive into the research topics of 'Automated vessel edge detection in velocity-encoded cine-MR (VEC-MR) flow measurements: A retrospective evaluation in critically ill patients'. Together they form a unique fingerprint.

Cite this