Towards Automated Extraction of Vessel Boundaries in Ultrasonic Blood Flow Images, by Means of Edge Detection

  • T. Loupas
Part of the Acoustical Imaging book series (ACIM, volume 22)


Because of its non-invasive and real-time nature, ultrasonic colour Doppler blood flow imaging has become an important clinical tool in the assessment of vascular disease1. Until now, the morphological and haemodynamic information provided by this modality is used purely for qualitative visual interpretation. The key to the development and widespread acceptance of quantitative applications of ultrasonic blood flow imaging lies in the ability to extract vessel boundaries in a robust and automated manner. The automated extraction of vessel boundaries is a prerequisite for a variety of geometrical analysis (diameter, area, volume, and beam-vessel angle measurements) and advanced image processing operations (3D reconstruction of complex vascular structures, registration of image sequences, vessel identification and segmentation). In direct analogy with quantitative X-ray angiography2, these operations are primarily applicable in the objective assessment of vascular morphology, but they are also relevant in the context of emerging techniques such as quantitation of tumour/organ vascularity3 and contrast-based Doppler intensitometry4. Colour Doppler scanners carry out a simple form of blood/tissue segmentation, so that the colour-coded flow map and the grey-scale anatomical image can be combined in the same display, which is primarily based on Doppler power thresholding5. Briefly, a pixel is classified as flow or tissue depending on whether its corresponding Doppler power is greater or lower, respectively, than a threshold value which is user-adjusted by means of the colour Doppler “gain” (or “level”) control. This criterion relies on the premise that, after clutter suppression, the Doppler signal associated with slowly moving or stationary soft tissue consists predominantly of electronic noise whose power is normally weaker than the power of the Doppler signal from moving blood. Although blood/tissue segmentation by means of power-thresholding is adequate for display purposes, it tends to be unreliable for quantitative applications. For example, the finite extent of the sample volume causes the Doppler power of pixels lying between adjacent vessels to be well above the noise floor, resulting in poor vessel separation. Also, the blurred nature of the flow-noise transitions, due to a combination of the transmitted pulse,s shape and clutter filter, implies that the flow-region dimensions are threshold-dependent (see Figure 1) and hence inappropriate from a quantitative point of view.


Edge Detection Doppler Power Doppler Power Image False Edge Stationary Soft Tissue 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    P. Lanzer and A. P. Yoganathan. “Vascular Imaging by Color Doppler and Magnetic Resonance,” Springer-Verlag, Berlin (1991).CrossRefGoogle Scholar
  2. 2.
    S. R. Fleagle, M. R. Johnson, C. J. Wilbricht, D. J. Skorton, R. W. Wilson, C. W. White, M. L. Marcus, and S. M. Collins, Automated analysis of coronary arterial morphology in cineangiograms: geometric and physiologic validation in humans, IEEE Trans. Med. Imaging 8:387 (1989).CrossRefGoogle Scholar
  3. 3.
    D. O. Cosgrove, J. C. Bamber, J. B. Davey, J. A. McKinna, and H. D. Sinnett, Color Doppler signals from breast tumors, Radiology 176:175 (1990).Google Scholar
  4. 4.
    N. Nanda and R. Schlief. “Advances in Echo Imaging Using Contrast Enhancement,” Kluwer Academic, Amsterdam (1993).Google Scholar
  5. 5.
    T. Loupas, R. B. Peterson, and R. W. Gill, Experimental evaluation of velocity and power estimation for ultrasound blood flow imaging, by means of a two-dimensional autocorrelation approach, IEEE Trans. Ultrason. Ferroelec. Freq. Control 42:689 (1995).CrossRefGoogle Scholar
  6. 6.
    D. Marr, and E. Hildreth, Theory of edge detection, Proc. Royal Soc. London B207:187 (1980).ADSCrossRefGoogle Scholar
  7. 7.
    J. Canny, A computational approach to edge detection, IEEE Trans. Pattern Anal. Machine Intell. 8:679 (1986).CrossRefGoogle Scholar
  8. 8.
    M. Shah, A. Sood, and R. Jain, Pulse and staircase edge models, Comput. Vision Graphics Image Process. 34:321 (1986).CrossRefGoogle Scholar
  9. 9.
    P. Perona, and J. Malik, Scale-space and edge detection using anisotropic diffusion, IEEE Trans. Pattern Anal. Machine Intell. 12:629(1990).CrossRefGoogle Scholar
  10. 10.
    C. Kasai, K. Namekawa, A. Koyano, and R. Omoto, Real-time two-dimensional blood flow imaging using an autocorrelation technique, IEEE Trans. Sonics Ultrason. 32:458 (1985).CrossRefGoogle Scholar
  11. 11.
    J. M. Rubin, R. O. Bude, P. L. Carson, R. L. Bree, and R. S. Adler, Power Doppler US: a potentially useful alternative to mean frequency-based color Doppler US, Radiology 190:853 (1994).Google Scholar
  12. 12.
    A. A. Farag, and E. J. Delp, Edge linking by sequential search, Pattern Recogn. 28:611 (1995).CrossRefGoogle Scholar
  13. 13.
    M. Kas, A. Witkin, and D. Terzopoulos, Snakes: active contour models, Int. J. Comput. Vision 1:321 (1988).CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 1996

Authors and Affiliations

  • T. Loupas
    • 1
  1. 1.Ultrasonics LaboratoryCSIRO Division of RadiophysicsChatswoodAustralia

Personalised recommendations