Name
Abstract | ||
|---|---|---|
We propose a new algorithm for non-assisted segmentation of possibly clustered nuclei from histological images. We use elliptic shapes as parametric models to represent the nuclei contours and fit the parameters using the information present in the gray level intensity image and in the derived gradient image. Multiple seeds for each closed contour are found by ultimate erosion of an estimated edge image, resulting in an number of seeds generally larger than the number of nuclei. Our algorithm, called segmentation of nuclei by ellipse fitting (SNEF), constructs several candidate contours for each seed by fitting ellipses to selected subsets of edge pixels. In the end the algorithm selects the contours to be declared nuclei by comparing the values of a suitably chosen goodness of fit criterion. The proposed algorithm produces segmentations in agreement with an expert pathologist. |
Year | Venue | Keywords |
|---|---|---|
2010 | Aalborg | biological tissues,image colour analysis,image segmentation,medical image processing,pattern clustering,snef,cell nuclei segmentation,edge image estimation,ellipse fitting,gradient imaging,gray level intensity imaging,histological imaging,multiple seed,nonassisted segmentation algorithm,parametric model,segmentation of nuclei by ellipse fitting,ultimate erosion,shape,clustering algorithms,silicon |
Field | DocType | ISSN |
Scale-space segmentation,Parametric model,Pattern recognition,Segmentation,Segmentation-based object categorization,Image segmentation,Artificial intelligence,Pixel,Ellipse,Cluster analysis,Mathematics | Conference | 2219-5491 |
Citations | PageRank | References |
4 | 0.52 | 4 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jenni Hukkanen | 1 | 4 | 0.86 |
| Andrea Hategan | 2 | 5 | 1.57 |
| Edmond Sabo | 3 | 7 | 1.94 |
| Ioan Tabus | 4 | 276 | 38.23 |