igp ncrr image and geometry processing highlights of ongoing work geometry processing shape...

IGP

NCRR

Image and Geometry Processing

Highlights of ongoing work•Geometry processing•Shape analysis•Visualization/segmentation

IGP

NCRR

Geometry Processing/Modeling

Volumes Label Maps

Corre

ct T

opol

ogy

Surface Meshes

Body-Fitted Meshes

SegmentationTools

VolumetricTopology Filtering

Antialiasing

Meshing: Adaptive, Geometrically Regular

Integrate 3rd Party Packages (Tets)

IGP

NCRR

Shape AnalysisCorrespondencesGeodesics on Solid Shapes

IGP

NCRR

Shape Correspondence ProblemShape metrics that rely on

correspondence

Correspondence depends on context of ensemble

IGP

NCRR

Particle Systems for Adaptive Surface Sampling

Meyer et al., Under review, IEEE Vis 2006

IGP

NCRR

A Particle-Based ApproachMinimize shape and ensemble entropy

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Cates et al., To appear MICCAI 2006 Workshop on Foundations of Computational Anatomy

IGP

NCRR

Statistical Shape Analysis

QuickTime™ and aMPEG-4 Video decompressor

are needed to see this picture.

IGP

NCRR

Geodesics of Solid Shapes

Fletcher and Whitaker, To appear MICCAI 2006 Workshop on Foundations of Computational Anatomy

IGP

NCRR

Segmentation Research

Neighborhood statistics•Manifolds in high-dimensional

spaces• Filtering, compression,

segmentation, visualizationStrategy•Nonparametric density

estimation• Engineering issues

IGP

NCRR

MRI Head SegmentationMICCAI 2005, MedIA to appear

MRI Input GM, WM, CSF Seg. Comparison: SOTA–EM w/MRFs & Atlas(Leemput et al.)

GM Classification Performance vs Noise Level

Collab: Makeig-Worrell, Wolters, Warfield, McIntyre

85

87

89

91

93

95

97

0% 1% 3% 5% 7% 9%

Proposed Leemput

IGP

NCRR

Cell/Texture SegmentationNonparametric density estimation – image neighorhoods

• TexturePartition image to reduce in-class entropyRandom initializationFast (nonlocal) level-set method

Collab: NCMIR, Marc

IGP

NCRR

IGP Research PlansContinued development of

meshing capabilitiesShape analysis development

and applicationApplications of neighborhood

statistics to visualization of complex data