Edge Detection in Ultrasonic Speckle

In this project, the problem of detecting boundaries in ultrasound speckle imagery is considered. For physiological reasons, we argue that boundaries between tissue layers appear as lines in ultrasound B-scans, and approach the boundary detection problem as one of detecting lines of unknown orientation. We define a set of "sticks," short line segments of variable orientation which can locally approximate the boundaries. Using the physical principles which account for the speckle phenomenon, we derive the optimal detector for sticks of unknown orientation in fully developed speckle, and compare the optimal detector to several sub-optimal detection rules which are more computationally efficient.

The basic technique is then extended in a number of way to improve its performance. We investigate the effect of varying the size and shape of the sticks, and show that these variations affect the performance of th algorithm in very fundamental ways, for example, by making it more or less sensitive to thinner to more tightly curving boundaries. We also present a means of improving performance by estimating the distribution function of the orientation of the line passing through each point. Next, we show that images can be "stained" for easier visual interpretation by applying to each pixel a false color whose hue is related to the orientation of the most prominent line segment at that point. Finally, an analysis is given of the boundary detection approaches in radio frequency ultrasound.

A typical image before (above) and after (below) processing with sticks operators. The processed
image is clearly enhanced, showing even thin lines without smearing, and rejecting background speckle,
with typically appears as spots rather than lines.

A testing image before (above) and after (below) processing with Signal Dependent Stick's Algorithm.

References

1) Hildreth, E. and D. Marr, "Theory of Edge Detection", Proceedings of Royal Society of London, 207, p. 187-217, 1980.

2) A. C. Bovik, "On Detecting Edges in Speckle Imagery", IEEE Trans on Signal Processing, vol. 36, no. 10, p. 1618-1627, October 1988.

3) R.O. Duda and P.E. Hart, "Use of the Hough Transform to Detect Lines and Curves in Pictures", Communications of the ACM, vol. 15, no. 1, p. 11-15, January 1972.

4) Y.K. Lee and W.T. Rhodes, "Scale and Rotation Invariant Pattern Recognition by a Rotating Kernal Min-Max Transformation", Optical Information Processing Systems and Architecture II, Franklin, T. Luke, Ed., Proc. SPIE 1347, p. 146-155,1990.

5) R.N. Czerwinski, D.L. Jones, and W.D. O'Brien, Jr., "An Approach to Boundary Detection in Ultrasound Imaging", 1993 IEEE Ultrasonics Symposium Proceedings, Baltimore, MD, November 1993.

6) R.N. Czerwinski, D.L. Jones, and W.D. O'Brien, Jr., "Edge Detection in Ultrasound Speckle Noise", Proceedings of the IEEE International Conference on Image Processing - ICIP '94, Austin, TX, November 1994.

7) R.N. Czerwinski, D.L. Jones, and W.D. O'Brien, Jr., "Ultrasound speckle Reduction by Directional Median Filtering", Proceedings of the IEEE International Conference on Image Processing - ICIP '95, Washington, DC, October 1995.

8) R.N. Czerwinski, D.L. Jones, and W.D. O'Brien, Jr., "Decision Directed Line Detection with Application to Medical Ultrasound", Proceedings of the 1996 SPIE International Symposium on Medical Imaging, Newport Beach, CA, February 1996.

9) Q. Tian, X. Li and N.M. Bilgutay, "Multiple Target Detection Using Split Spectrum Processing and Group Delay Moving Entropy", IEEE Trans on UFFC, vol. 42, no. 6, p. 1075-1086, November 1995.

10) Q. Tian and N.M. Bilgutay, "Statistical Analysis of Split Spectrum Processing for Multiple Target Detection", IEEE Trans on UFFC, vol. 45, no. 1, p. 251-256, January 1998.