Tissue image reconstruction: Localization of nuclei markers and segmentation based on deformable models
| Authors | |
|---|---|
| Year of publication | 2003 |
| Type | Article in Proceedings |
| Conference | Biophysics of the Genome |
| MU Faculty or unit | |
| Citation | |
| Field | Informatics |
| Keywords | image analysis |
| Description | Tissue segmentation has not been safisfactory solved yet. Several recent papers [3,4] brought some new pieces of knowledge into this area. However there is a lot of open questions. Improving the degree of automation of tissue segmentation is a great challenge. A new 3-step semiautomatic method was developed for tissue segmentation. The method is suitable for specifically-shaped cells. The first step includes searching for cells markers, i. e. the approximate center of each cell is localized. The localization is based on careful analysis of cell boundaries and on the assumption that the cells are sphere-like objects. The main contribution of the method is the possibility to find the cells markers without choosing the particular cells by hand. In the next step the surface of each cell is reconstructed. The procedure is based on the method for spherical object reconstruction presented in [7]. That method uses star-shaped simplex mesh to reconstruct the cell surface. The method was partially changed and was adapted to be more suitable for our purposes. First of all the problem of getting stuck in local minima had to be solved. Therefore another type of simplex mesh called dual simplex mesh was used. In addition the deformation process was sped up. The final step concerns evaluation of the results: both of the first two steps are nearly automatic, therefore the quality of their results should be measured. |
| Related projects: |