Accuracy Evaluation of Differents Centerlines Approximation of Blood Vessels

Abstract

Accurate determination of the central vessel axis is a prerequisite for automated arteries diseases visualization and quantification. In this paper we present an evaluation of different methods used to approximate the centerline of the vessel in a phantom simulating the peripheral arteries. Six algorithms were used to determine the centerline of a synthetic peripheral arterial vessel. They are based on: ray casting technique using thresholds and maximum gradient-like stop criterion, pixel motion estimation between successive images called block matching, center of gravity and shape based segmentation. The Randomized Hough Transform and ellipse fitting using Lagrange Multiplier have been used as shape based segmentation techniques, fitting an elliptical shape to a set of points. The synthetic data simulate the peripheral arterial tree (aorta-to-pedal). The vessel diameter changes along the z-axis from about 0.7 to about 23 voxels. The data dimension is 256x256x768 with voxel size 0.5x0.5x0.5mm. In this data set the centerline is known and an estimation of the error is calculated in order to determine how precise a given method is and to classify it accordingly.

Keywords:Centerline Detection, Vessel Segmentation, Medical Visualization

@InProceedings{ALCVISSYM04,
  author = "A. {La Cruz}",
  title = "{A}ccuracy {E}valuation of {D}ifferent 
          {C}enterline {A}pproximations of {B}lood 
		  {V}essels",
  booktitle = {Data Visualization 2004, Visualization 
          Symposium Proceedings},
  pages = {115--120},
  year = {2004},
  Month = {May},
  publisher = {EUROGRAPHICS/IEEE TCVG},
}

Abstract

Download full paper

Figures in the paper

BibTeX Entry