The purpose of this study is the contour extraction of a region of interest in ultrasonic images. I applied the anisotropic diffusion algorithm to preprocess images, then an active contour model using a gradient vector flow was employed. In the end, the contour of a lesion area of the ultrasonic images were extracted.

--------------------Author by Yao Zhang

• Main.m -click to run the whole problem
• AnisotropicDiffusion.m -Used for smooth image preserving the edges in the image at same time.
• GVF.m -Compute the gradient vector flow.
• imdisp.m -scale the dynamic range of an image and display it.
• snakedeform.m -In this function, the initial contour of Active Contour Model(Snake) will be deformed in the given external force field.
• snakedisp.m -Display the snake model contour
• snakeindex.m -Create index for adaptive interpolating the snake
• snakeinit.m -Implement Canny Operator to initialize initial contour line for Active Contour Model
• snakeinterp.m -interpolate the snake adaptively

So we can actually see that images noise are removed and the edges of object in the images are preserved.

There are initial contour line found by Canny Operator and corresponding final segmentation result.

1. First Example

• Final result

2. Second Example

• Final result

3. Third Example

• Final result

• Perona P, Malik J. Scale-space and edge detection using anisotropic diffusion[J]. IEEE Transactions on pattern analysis and machine intelligence, 1990, 12(7): 629-639.
• Kass M, Witkin A, Terzopoulos D. Snakes: Active contour models[J]. International journal of computer vision, 1988, 1(4): 321-331.
• Yang-xu C. Snakes, shapes, and gradient vector flow[C]//Inter-national Conference on Image Processing. 2002, 9(2): 17-820.
• Gradient Vector Flow (GVF) Active Contour Toolbox by Chenyang Xu and Jerry Prince