Edge-based image segmentation: machine learning from examples

We report a method for the design of optimal edge based image segmentation systems in which the criterion of optimality is automatically determined by learning from border tracing examples. The border features employed in the designed method are selected from a predefined global set using radial-basis neural networks. The method was validated in intracardiac, intravascular, and ovarian ultrasound images. The achieved performance was comparable to that of our previously reported single-purpose border detection methods (Sonka et al. (1995). Our approach facilitates development of general multipurpose image segmentation systems that can be trained for different types of image segmentation applications.