Graphical Models and Image Processing
Scale-based fuzzy connected image segmentation: theory, algorithms, and validation
Computer Vision and Image Understanding - Special issue on analysis of volumetric image
Computer Vision and Image Understanding
| Hi-index | 0.00 |
A framework based on atlas registration is proposed for automatic identification and segmentation of abdominal organs. The VIP-Man atlas is adopted to guide the whole process. The atlas was registered onto the subject through global registration and organ registration. In global registration, an affine transformation was found to eliminate the global differences between the atlas and the subject, using normalized mutual information as the similarity measure. In organ registration, organs of interest were registered respectively to achieve better alignments. An original similarity measure was proposed in organ registration. The registered atlas can be viewed as an initial segmentation of the subject, and make organs of interest identified. As an application of the registered atlas, novel methods were designed to estimate necessary parameters for fuzzy connectedness (FC) segmentation. Manual intervention was avoided, and thus to increase the automation degree of the method. This atlas-based method was tested on abdominal CT images of Chinese patients. Experimental results indicated the validity of the method for both male and female subjects of different ages.