Conference Paper

A Knowledge-based Segmentation Method Integrating both Region and Boundary Information of Medical Images.

DOI: 10.1109/BMEI.2008.64 Conference: Proceedings of the 2008 International Conference on BioMedical Engineering and Informatics, BMEI 2008, May 28-30, 2008, Sanya, Hainan, China - Volume 1
Source: DBLP


In this article, the author proposed a hybrid segmentation method which integrates region, boundary and priori knowledge information of medical images. The basic algorithm of this method is level set active contours. The speed function is initialized according to the gradient of the image, and is modified according to statistical characteristic of the segmented regions as the curve evolves. To make the curve stop accurately at the boundary of the object, an energy function is constructed by improving Chan-Vese model. The priori knowledge of the region of interest (ROI) is also integrated into this energy function. The experiment data consists of both simulated images and real clinical images. Precision, accuracy and efficiency are considered in evaluating this method. The evaluation result shows that this method is robust, accurate and has high performance, especially when the boundary is weak or dotted.

0 Reads
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, based on the ideas of image texture analysis and motion tracking, we present a new method for left ventricular (LV) strain analysis from cine magnetic resonance images(MRI). First, the marking points are extracted with the Harris corner detector on or close to the endocardium and epicardium. Second, using the tree-structured wavelet algorithm, the corresponding marking points are matched automatically and the sparse displacement field is calculated precisely between the consecutive frames. Next, applying radial point interpolation with polynomial basis functions, which is used in the meshfree method, the dense displacement field is interpolated. Finally, LV strains is analyzed. The experimental results show that our method is feasible and effective.
    No preview · Conference Paper · Jan 2009
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, integrating boundary and region information of medical images, we propose a novel hybrid segmentation method based on level set. The main contributions of this paper are to modify the velocity function for the boundary-based level set method, and to design a novel energy function as a stopping criterion. This velocity function is modified according to the statistical characteristics of the segmented regions during the evolution so that the medical images with weak boundary and concave region can be segmented. The stopping criterion depends on not only the boundary information of the image but also the statistical characteristics of the segmented regions, which can overcome the over-segmentation effectively. Furthermore, our method forces the level set function close to a signed distance function, therefore, eliminates the complex re-initialization procedure and reduces the side effects of re-initialization. Experimental results for real clinical images show the effectiveness of our method.
    No preview · Article · Oct 2009
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a improved Markov Random Field (MRF) segmentation model, which integrates region, priori knowledge and boundary information of the image, for segmenting left ventricle (LV) boundary from cardiac MR image. The proposed model incorporates geometry shape boundary information, and improves the objective function of traditional MRF model. Furthermore, Chaotic Simulated Annealing (CSA) algorithm is introduced to solve the MRF model for the first time. Since CSA algorithm introduces chaos ergodicity mechanism, it can take advantage of Chaos Algorithm (COA) and Simulated Annealing (SA) algorithm in the search process. CSA algorithm can not only avoid the limitations of mathematical optimization methods, but also greatly enhance the speed of global optimization. Experiments on clinical cardiac MR images show that the improved MRF model has high performance on segmenting LV boundary. The evaluation results illustrate that this model is robust, accurate and efficient, especially for the weak boundary and concave region .
    No preview · Conference Paper · Nov 2009
Show more