Technical Section: Variational implicit surface meshing
Computers and Graphics
Variational level-set with gaussian shape model for cell segmentation
ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Feature-based registration of confocal fluorescence endomicroscopy images
ICIP'09 Proceedings of the 16th IEEE international conference on Image processing
Geometry-driven visualization of microscopic structures in biology
EuroVis'08 Proceedings of the 10th Joint Eurographics / IEEE - VGTC conference on Visualization
Hi-index | 0.00 |
Early efforts in the visualization of microscopic biological structures have been impeded by the lack of a rigorous cellular segmentation approach, insufficient slicing resolution and slicing deformations associated with serial-section histology volumes. We develop algorithms that address these challenges. In this paper, geodesic active contours using shape priors is employed to obtain initial segmentations of salient cellular structures. Overlapping cells are resolved by imposing a Voronoi-like tessellation of the image space optimized by a Bayesian probability framework. Intermediate slices are introduced between images to account for the insufficient slicing resolution. Results of the cellular segmentation step are used in conjunction with a cell shape model to interpolate the 3D cellular locations and shapes onto the adjacent slices thereby enhancing the expressivity and utility of the resulting visualizations. Our methods are applied in a case-study involving the 3D visualization of the epithelial cell lining and lobules in mouse mammary ducts.