Abstract | ||
---|---|---|
Original medical images are often inadequate for clinical diagnosis. Certain prior information can be used as an important basis for disease diagnosis and prevention. In this study, an oriented grouping-constrained spectral clustering method, OGCSC, is proposed to deal with medical image segmentation problems. OGCSC propagates the group information from the affinity matrix and subdivides the group information into two constraints. By adopting the normalized framework, OGCSC can be transformed into normalized spectral clustering. The solution of OGSCSC can be viewed as a generalized eigenvalue problem that can be solved using eigenvalue decomposition techniques. The significance of our work is that the use of group information and constraints information to analyse image data can greatly enhance the results achieved using the clustering segmentation method. The empirical experimental results reveal that the proposed method achieves robust and effective performance for medical image segmentation. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1007/s00530-019-00626-8 | Multimedia Systems |
Keywords | Field | DocType |
Medical image segmentation, Spectral clustering, Semi-supervised clustering, Constrained information | Spectral clustering,Normalization (statistics),Pattern recognition,Computer science,Medical imaging,Segmentation,Real-time computing,Image segmentation,Artificial intelligence,Eigendecomposition of a matrix,Clinical diagnosis,Cluster analysis | Journal |
Volume | Issue | ISSN |
26 | 1 | 0942-4962 |
Citations | PageRank | References |
2 | 0.37 | 20 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Kaijian Xia | 1 | 2 | 0.70 |
Xiaoqing Gu | 2 | 44 | 9.30 |
Yudong Zhang | 3 | 251 | 25.00 |