Heuristic Linking Models in Multiscale Image Segmentation

Koster, André S. E.; Vincken, Koen L.; Graaf, Cornelis N. de; Zander, Olaf C.; Viergever, Max A.

doi:10.1006/cviu.1996.0490

Cited by 22 publications

(12 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, there has been an attempt to fuse gradient dependent diffusion (GDD)based edge information with hyperstack-based linking approaches for 3-D cortical segmentation (see the three Ph.D.s from Utrecht University: Koster [507], Vincken [508], Niessen [509] and the four recent papers by Koster et al [510], Vincken et al [511] and Niessen et al [512], [513]). We will summarize their ideas in cortical segmentation.…”

Section: Regional Hyperstack: Fusion Of Edge-diffusion With Region-limentioning

confidence: 99%

Advanced Algorithmic Approaches to Medical Image Segmentation

Suri¹,

Setarehdan²,

Singh³

2002

View full text Add to dashboard Cite

Section: Regional Hyperstack: Fusion Of Edge-diffusion With Region-limentioning

confidence: 99%

Advanced Algorithmic Approaches to Medical Image Segmentation

Suri¹,

Setarehdan²,

Singh³

2002

View full text Add to dashboard Cite

“…The flavor of our paper is close to the comparison of diffusion schemes for segmentation performed on the hyper-stack [33,18,24]. The major differences are that the hyper-stack is based on isophote linking and that it constitutes an automated segmentation algorithm.…”

Section: Introductionmentioning

confidence: 96%

Non-linear Diffusion for Interactive Multi-scale Watershed Segmentation

Dam

Nielsen

2000

Medical Image Computing and Computer-Assisted Intervention – MICCAI 2000

View full text Add to dashboard Cite

Abstract. Multi-scale watersheds have proven useful for interactive segmentation of 3D medical images. For simpler segmentation tasks, the speed up compared to manual segmentation is more than one order of magnitude. Even where the image evidence does not very strongly support the task, the interactive watershed segmentation provides a speed up of a factor two. In this paper we evaluate a broad class of non-linear diffusion schemes for the purpose of interactively segmenting gray and white matter of the brain in T1-weighted MR images. Through a new scheme GAN, we show that diffusion similar to the nonlinear PeronaMalik scheme is superior to the other evaluated diffusion schemes. This provides a speed up factor of two compared to the linear diffusion scheme.

show abstract

“…The hyperstack (for a more detailed description and extensions see [12,131) is a linking model based segmentation technique. It USES the luminance in adjacent levels to relate points.…”

Section: Hyperstackmentioning

confidence: 99%

“…1 teria are the luminance value and proximity; a parent is selected in a scale dependent search region based on intensity difference. Also an affection strength is assigned to the linkages based on heuristic and statistical features (see [12] for details). The linking is a bottom-up process such that only parents that have been linked to are considered children in the next linking step.…”

Section: Hyperstackmentioning

confidence: 99%

Three-dimensional MR brain segmentation

Niessen

Vincken²,

Weickert³

et al.

Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271)

View full text Add to dashboard Cite

In M R brain images, segmentation using intensity values is severely limited owing to field inhomogeneities, susceptibility artifacts and partial volume effects. Edge based segmentation methods suffer f r o m spurious edges and gaps in boundaries. A ,method is presented which combines the advantages of edge based and region based segmentation. Farst a multiscale image representation is constructed which favors intratissue diffusion over inter-tissue diffusion by exploiting local contrast. Subsequently a multiscale linking model (the hyperstack) is used t o group voxels into a number of segments. This facilitates segmentation of grey matter, white matter and cerebrospinal fluid with minimal user interaction. Using a supervised segmentatzon technique and M R simulations of a brain phant o m as validation it is shown that the errors are in the order of or smaller than reported in literature.

show abstract

Heuristic Linking Models in Multiscale Image Segmentation

Cited by 22 publications

References 16 publications

Advanced Algorithmic Approaches to Medical Image Segmentation

Advanced Algorithmic Approaches to Medical Image Segmentation

Non-linear Diffusion for Interactive Multi-scale Watershed Segmentation

Three-dimensional MR brain segmentation

Contact Info

Product

Resources

About