An efficient convex optimization approach to 3D prostate MRI segmentation with generic star shape prior

Jing Yuan, Wu Qiu, Eranga Ukwatta, Martin Rajchl, Yue Sun, Aaron Fenster

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

4 Downloads (Pure)

Abstract

In this work, we propose a novel global optimization-based contour evolution approach for the segmentation of 3D prostate magnetic resonance (MR) images, which incorporates histogram-matching and a novel variational formulation of a generic star shape prior. Our method overcomes the existing challenges of segmenting 3D prostate MRI: heterogeneous intensity distributions and a wide variety of prostate shape appearances. We introduce a novel convex relaxation-based method to evolve a contour to its globally optimal position during each discrete time frame. The proposed generic star shape prior provides robustness to the segmentation when the image suffer from poor quality, noise, and artifacts. Our approach provides a fully time implicit scheme to contour evolution, which allows a large time step-size to accelerate the speed of convergence. Moreover, a new continuous max-flow formulation is pro posed which is dual to the convex relaxation formulation, obtains global optimality of contour evolution, and is implemented in a GPU.
Original languageEnglish
Title of host publicationProc. Med Image Comput.-Assisted Intervent. Conf. Prostate Segment. Challenge 2012
Pages82-89
Number of pages8
Publication statusPublished - 2012
Externally publishedYes

Fingerprint

Dive into the research topics of 'An efficient convex optimization approach to 3D prostate MRI segmentation with generic star shape prior'. Together they form a unique fingerprint.

Cite this