A probability model-based level set method for biomedical image segmentation

Level set methods are useful tools for medical image segmentation. In this paper, a novel image segmentation technique was developed that combines region statistical information with the level set method. The classical level set methods depend mainly on local edge-based image features to guide the convergence of the contour. This makes the method sensitive to noise and the initial estimate. The method has two features. The first is that it uses obtain region-based information by a mixture model. The second feature is that we combine region statistical information with curvature-based regularization penalizing the length of curve. The method is useful for a large variety of segmentation problems. We present some preliminary experimental results using synthetic images and, magnetic resonance (MR), ultrasound (US), computed tomography (CT) images to demonstrate our methods. The experimental results show that by incorporating region statistical information into the level set framework, an accurate and robust segmentation can be achieved.