Inhalt des Dokuments
Preprint 33-2007
A three stepped coordinated Level Set Segmentation Method for Identifying atherosclerotic plaques on MR-images
Author(s) :
Oliver Gloger
,
Matthias Ehrhardt
,
Olaf Hellwich
,
Eike Nagel
Preprint series of the Institute of Mathematics, Technische Universität Berlin
Preprint 33-2007
MSC 2000
- 92B05 General biology and biomathematics
-
65M06 Finite difference methods
Abstract :
In this work we propose an adapted level set segmentation technique for the recognition of atherosclerotic plaque tissue on magnetic resonance images. The segmentation technique is subdivided into three steps whereas the result of every phase serves as additional prior knowledge for the next level set step. By incorporating special knowledge derived from the images into the level set equation we guide the level set towards the desired plaque patterns. By analyzing and applying carefully all disposable channel information we are capable to border the vessel walls on the images and after that approach an enclosing level set which circumvents the plaque patterns from healthy media tissue. We support the plaque detection with so-called canny edges and locally weighted intensity information for
conspicuous plaque patterns. We justify the introduction of a maximal shrinking distance of the 3rd level set in the vessel wall for supporting segmentation and show very promising results of this method in special figures during the level set propagation.
By extending the level set segmentation with our concepts of
locally weighted intensity information and the maximal shrinking distance we teach the level sets to segment the dangerous plaque patterns very exactly. Furthermore we are capable to emphasize and segment plaque patterns which are difficult to recognize even for the human observer.
Keywords :
level set segmentation, active contours, medical image segmentation, anisotropic diffusion, atherosclerotic plaques, canny edges
Notes :
submitted to: Mathematical Modelling of Natural Phenomena (Special Issue of Medical Imaging, December 2007)