A modified snake method based on the novel idea of coordinate regeneration is presented and is tested on an object with complex concavities and on retinal images for locating the boundaries of optic discs, where the conventional snake methods fail. We have demonstrated that the use of conventional snake method with our proposed coordinate regeneration technique gives ultimate solution for finding the boundaries of complex objects. The proposed method requires a Gaussian blur of the object with a large kernel so that the snake can be initialised away from the object boundaries. In the second and third steps the blurring kernel size is reduced so that exact boundaries can be located. Coordinate regeneration is applied at each step which ultimately converges the snake (active contour) to exact boundaries. For complex objects like optic discs in retinal images, vessels act as snake distracters and some preimage processing is required before the proposed technique is applied. We are demonstrating this technique to find the boundary of optic discs in retinal images. In principle, this technique can be extended to find the boundary of any object in other modalities of medical imaging. Simulation results are presented to support the idea. 1. Introduction In this section, we are introducing a modified form of snake method used for automatic tracking of boundary of an object in a medical image. We are considering retinal images as medical images and optic disc as an object of interest. The first snake model was introduced by Kass et al. [1]. Mendels et al. [2] were the persons who applied this technique for the first time to track the boundary of the optic disc in retinal images. The concept of gradient vector flow (GVF) snake was introduced to extend the range of initialisation of snakes [3] to make the snake method more effective referred to as GVF snakes. We have made some changes to the normal snake method by introducing a new concept of coordinate regeneration on the curve and made it more effective. The main advantage of coordinate regeneration is to extend the range for initialisation of snakes around or within a boundary and to come out with a more precise boundary data. The algorithm based on coordinate regeneration is much faster and more robust when combined with normal snake method. Later in this paper, we will establish that the concept of coordinate regeneration alleviates the need of GVF snakes which has been a hot topic for more than a decade. For completeness, we are giving a brief introduction to the theory of normal and GVF-snake models, and
References
[1]
M. Kass, A. Witkin, and D. Terzopoulos, “Snakes: active contour models,” International Journal of Computer Vision, vol. 1, no. 4, pp. 321–331, 1988.
[2]
F. Mendels, C. Heneghan, and J. Thiran, “Identification of optic disc boundary in retinal images using active contours,” in Proceedings of Irish Machine Vision and Image Processing Conference (IMVIP '99), pp. 103–115, 1999.
[3]
J. L. Prince and C. Xu, “A new external force model for snakes,” in Proceedings of the Image and Multidimensional Signal Processing Workshop, pp. 30–31, 1996.
[4]
D. Terzopoulos and K. Fleischer, “Deformable models,” The Visual Computer, vol. 4, no. 6, pp. 306–331, 1988.
[5]
T. McInerney and D. Terzopoulos, “A dynamic finite element surface model for segmentation and tracking in multidimensional medical images with application to cardiac 4D image analysis,” Computerized Medical Imaging and Graphics, vol. 19, no. 1, pp. 69–83, 1995.
[6]
F. Leymarie and M. D. Levine, “Tracking deformable objects in the plane using an active contour model,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 15, no. 6, pp. 617–634, 1993.
[7]
R. ？urikovi？, K. Kaneda, and H. Yamashita, “Dynamic contour: a texture approach and contour operations,” The Visual Computer, vol. 11, no. 6, pp. 277–289, 1995.
[8]
D. Terzopoulos and R. Szeliski, “Tracking with kalman snakes,” in Active Vision, Artificial Intelligence, A. Blake and A. Yuille, Eds., pp. 3–20, The MIT Press, Cambridge, Mass, USA, 1992.
[9]
V. Caselles, F. Catte, T. Coll, and F. Dibbos, “A geometric model for active contours,” Numerische Mathematik, vol. 66, no. 1, pp. 1–31, 1993.
[10]
R. Malladi, J. A. Sethian, and B. C. Vemuri, “Shape modeling with front propagation: a level set approach,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 17, no. 2, pp. 158–175, 1995.
[11]
V. Caselles, R. Kimmel, and G. Sapiro, “Geodesic active contours,” in Proceedings of the 5th International Conference on Computer Vision (ICCV '95), pp. 694–699, June 1995.
[12]
L. D. Cohen, “On active contour models and balloons,” Graphical Models and Image Processing, vol. 53, no. 2, pp. 211–218, 1991.
[13]
L. D. Cohen and I. Cohen, “Finite-element methods for active contour models and balloons for 2-D and 3-D images,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 15, no. 11, pp. 1131–1147, 1993.
[14]
A. J. Abrantes and J. S. Marques, “A class of constrained clustering algorithms for object boundary extraction,” IEEE Transactions on Image Processing, vol. 5, no. 11, pp. 1507–1521, 1996.
[15]
C. A. Davatzikos and J. L. Prince, “Active contour model for mapping the cortex,” IEEE Transactions on Medical Imaging, vol. 14, no. 1, pp. 65–80, 1995.
[16]
C. Xu and J. L. Prince, “Snakes, shapes, and gradient vector flow,” IEEE Transactions on Image Processing, vol. 7, no. 3, pp. 359–369, 1998.
[17]
W. F. Ames, Numerical Methods for Partial Differential Equations, Computer Science and Scientific Computing, Academic Press, New York, NY, USA, 3rd edition, 1992.
