Fast Flow-Line-Based Analysis of Ultrasound Spectral and Vector Velocity Estimators

IEEE Trans Ultrason Ferroelectr Freq Control. 2019 Feb;66(2):372-381. doi: 10.1109/TUFFC.2018.2887398. Epub 2018 Dec 24.

Abstract

A new technique, termed FLUST (FlowLine Ultrasound Simulation Tool), is proposed as a computationally cheap alternative to simulations based on randomly positioned scatterers for the simulation of stationary blood velocity fields. In FLUST, the flow field is represented as a collection of flow lines. Point spread functions are first calculated at regularly spaced positions along the flow lines before realizations of single scatterers traversing the flow lines are generated using temporal interpolation. Several flow-line realizations are then generated by convolution with temporal noise filters, and finally, flow-field realizations are obtained by the summation of the individual flow-line realizations. Flow-field realizations produced by FLUST are shown to correspond well with conventional Field II simulations both quantitatively and qualitatively. The added value of FLUST is demonstrated by using the proposed simulation technique to obtain multiple realizations of realistic 3-D flow fields at a significantly reduced computational cost. This information is utilized for a performance assessment of different spectral and vector velocity estimators for carotid and coronary imaging applications. The computational load of FLUST does not increase substantially with the number of realizations or simulated frames, and for the examples shown, it is the fastest alternative when the total number of simulated frames exceeds 48. In the examples, the standard deviation and bias of the velocity estimators are calculated using 100 FLUST realizations, in which case the proposed method is two orders of magnitude faster than simulations based on random scatterer positions.

MeSH terms

  • Algorithms
  • Blood Flow Velocity / physiology*
  • Carotid Arteries / diagnostic imaging
  • Coronary Stenosis
  • Coronary Vessels / diagnostic imaging
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Models, Cardiovascular
  • Patient-Specific Modeling
  • Phantoms, Imaging
  • Ultrasonography / methods*